Webpdf

Best Practice in Inventory Management
Best Practice in Inventory Management, Third Edition offers a simple, entirely

jargon-free and yet comprehensive introduction to key aspects of inventory man-
agement. Good management of inventory enables companies to improve their cus-
tomer service, cash flow and profitability. This text outlines the basic techniques,
how and where to apply them, and provides advice to ensure they work to provide
the desired effect in practice.
With an unrivalled balance between qualitative and quantitative aspects of
inventory control, experienced consultant Tony Wild portrays the many ways in
which stock management is more nuanced than simple ‘number crunching’ and
mathematical modelling.
This long-awaited new edition has been substantially and thoroughly updated.
The product of decades of experience and expertise in the field, Best Practice in
Inventory Management, Third Edition provides students and professionals, even
those with no prior experience in the area, with an unbiased and honest picture of
what it takes to effectively manage stocks in a firm.
Tony Wild, PhD, is Managing Director of Dawson Berkeley & Partners Ltd, UK.
Tony is a leading expert on Inventory Management, developing and introducing
new ideas in large and small companies in their Supply Chains. He is also lecturing
at the University of Warwick. Tony would now like to pass on his secrets of success
to generations of students and practitioners, who can use them and take up this
exciting and challenging profession.
Best Practice in Inventory
Management
Third Edition
Tony Wild
First published 2018
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
711 Third Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2018 Tony Wild
The right of Tony Wild to be identified as author of this work has been
asserted by him in accordance with sections 77 and 78 of the Copyright,
Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in
any information storage or retrieval system, without permission in writing
from the publishers.
Trademark notice: Product or corporate names may be trademarks or
registered trademarks, and are used only for identification and explanation
without intent to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Wild, Antony, author.
Title: Best Practice in Inventory Management/Tony Wild.
Description: 3 Edition. | New York: Routledge, 2017. | Revised edition of
the author’s Best Practice in Inventory Management, [2002] | Includes
bibliographical references and index.
Identifiers: LCCN 2017023467 (print) | LCCN 2017025095 (ebook) |
ISBN 9781315231532 (eBook) | ISBN 9781138294424 (hbk.: alk. paper) |
ISBN 9781138308077 (pbk.: alk. paper) | ISBN 9781315231532 (ebk)
Subjects: LCSH: Inventory control.
Classification: LCC TS160 (ebook) | LCC TS160.W495 2018 (print) |
DDC 658.7/87–dc23
LC record available at https://lccn.loc.gov/2017023467
ISBN: 978-1-138-29442-4 (hbk)

ISBN: 978-1-138-30807-7 (pbk)
ISBN: 978-1-315-23153-2 (ebk)
Typeset in Times New Roman

by Sunrise Setting Ltd, Brixham, UK
Visit the companion website www.routledge.com/cw/Wild

Contents
List of figures ix
Preface xii
Introduction xiv
1 The basis of inventory control 1

1.A The role of inventory management 1
1.B Objectives for inventory control 4
1.C Profit through inventory management 7
1.D Reasons for the current stock 9
2 Customer service 16
2.A Customer relations 16
2.B Measuring availability 17
2.C Demand management 24
2.D Consuming forecast demand and supply lead time 28
3 Shaping inventory 33
3.A Using Pareto Analysis for control 33
3.B ABC analysis 36
3.C Stock cover 43
3.D Pareto stock balance 48
4 Practical methods for reducing

stockholding 51
4.A Approach to inventory reduction 51
4.B The reduction project 55
4.C Obsolete and excess stock 58
vi Contents
5 Management and control 62
5.A Where stock control fits into the organisation 62
5.B Responsibilities and targets 64
5.C Skills and systems 71
5.D Inventory valuation 75
6 Lean supply 80
6.A Lean supply philosophy 80
6.B Lean principles 84
6.C Implementing lean 89
6.D Operational benefits of lean supply 93
6.E Developing lean operations 96
7 Safety stocks 98
7.A Learning from history 98
7.B Normal demand patterns 101
7.C Evaluating safety stocks 106
8 Setting the right stock levels 113

8.A Simple assessment of review levels 113
8.B Managing lead times 117
8.C Supplier delivery frequency effects 122
8.D Target stock levels 124
9 Procurement 128
9.A The role of supply chain procurement 128
9.B The purchasing environment 131
9.C Single sourcing 132
9.D Supply partnerships 133
9.E Vendor appraisal 135
9.F Pricing methods 138
10 Delivery quantities 140

10.A Supply and suppliers 140
10.B Organising repetitive supply 141
10.C Order types 142
10.D Order quantities 143
10.E Delivery quantities 144
Contents vii
10.F Scheduling supply 148
10.G Supply co-ordination 153
10.H Purchasing processes 153
11 Forecasting demand 156

11.A Options for assessing demand 156
11.B Causes of forecasting inaccuracy 162
11.C Methods of improving forecasting 166
12 Historical forecasting techniques 169

12.A Basic forecasting techniques 169
12.B Moving average 171
12.C Exponentially weighted averages 173
12.D Improved values for mean absolute deviation 178
12.E Choosing the best forecast – focus forecasting 179
13 Improved forecasting methods 188

13.A More forecasting tools 188
13.B Forecasting for seasonal sales 194
14 Dependent demand 201

14.A Avoiding uncertainty 201
14.B Material requirements planning 203
14.C Master planning 210
14.D Master scheduling 210
15 Supply chain inventory management 217

15.A The basis of the lean supply chain 217
15.B Co-ordination 219
15.C Supply chain operations 222
15.D Replenishment techniques 226
15.E Managing distribution 230
15.F Recycling and reverse logistics 236
16 Meeting the challenges 238

16.A Review 238
16.B Collaboration 240
16.C Information structures 240
16.D Recipe for success 243
viii Contents
Questions and answers 246
Questions 246
Answers 257
Index 276
Figures
1.1 Which direction? 2

1.2 Cost element 8
2.1 Availability of a stock item 18
2.2 Alternative measurements of availability 20
2.3 On time or in full (OTIF) 22
2.4 Comparison of despatches for companies A and B 22
2.5 Despatch performance for companies A and B 22
2.6 Distribution of arrears 23
2.7 Managing arrears 23
2.8 Order processing options 27
2.9 Consuming the forecast 29
3.1 Pareto curve 34
3.2 ABC analysis 37
3.3 ABC inventory control 38
3.4 Example of Pareto Analysis 39
3.5 Classification by usage value 39
3.6 Summary of ABC analysis 39
3.7 The cumulative Pareto curve 40
3.8 Stock cover 44
3.9 Ranges of stock cover 46
3.10 Expected and actual stock profiles 47
3.11 Minimising stock value the easy way 47
4.1 Stock levels 52
4.2 Measuring obsolete and excess 58
4.3 Strategies avoiding surplus stock 59
4.4 Avoiding surplus inventory 59
4.5 Sequence for dealing with surplus stock 60
5.1 Evolution of supply chain operations 63
5.2 Management controls 66
5.3 Planning inventory value 70
5.4 Controlling total spend 71
5.5 Achieving better inventory control 72
5.6 KPI records for each stores 74
x Figures
5.7 Standard cost and variances 76
5.8 Alternative methods of determining unit stock value 78
6.1 Contrasts between conventional and JIT inventory control 81
6.2 Warehouse stockholding 81
6.3 Effect of throughput time on manufacturing inventory 82
6.4 Ideal situation for just-in-time operations 86
6.5 Continuous supply cycle times 87
6.6 Cyclic production 95
7.1 Demand history of item T15 99
7.2 Demand distribution 100
7.3 Normal distribution curve 100
7.4 Availability of a stock item 102
7.5 Demand history of item T26 102
7.6 Stock and availability 103
7.7 Summary of deviation measurement techniques 105
7.8 Customer service factors (assuming a normal distribution) 106
7.9 Stock record of item T26 108
7.10 Lead time multiplier for safety stock 109
7.11 Definition of excess and obsolete stock 111
8.1 Inventory level 114
8.2 Items with average demand of 31 114
8.3 Review levels 115
8.4 Varying the conditions 116
8.5 Safety stock and delivery frequency 122
8.6 Effect of supply batch size on availability 123
8.7 Safety stock correction for delivery quantity 124
8.8 Periodic review level 125
9.1 Kraljic’s Matrix 129
9.2 Supplier attitudes 130
9.3 Single sourcing advantages and disadvantages 134
9.4 Partner selection checklist 135
9.5 Classification of suppliers 137
9.6 Vendor appraisal 138
10.1 Supplier characteristics 141
10.2 Review of supply agreements 143
10.3 Pareto-based delivery quantities 145
10.4 Pareto order patterns for 1,000 stock lines 146
10.5 Classical evaluation of order quantity 147
10.6 The quantity required by the horizon cover calculation 148
10.7 Relative quantities 152
11.1 Product life cycle 160
11.2 Forecasting new products 161
11.3 Effect of amount of history on accuracy of forecast 166
12.1 Moving averages 171
12.2 Contribution from each period to the average 174
Figures xi
12.3 Exponential forecasting sensitivity 175
12.4 Exponential smoothing 176
12.5 Contributions of historical data to the forecast 177
12.6 Exponentially weighted MAD 180
12.7 Focus forecasting 181
12.8 Tracking signals 184
12.9 Trigg’s method 185
12.10 Forecasting with Trigg’s method 186
13.1 Improved forecasting through double exponential smoothing 189
13.2 Double exponential model 190
13.3 Double exponential smoothing process 191
13.4 Double exponential calculation 192
13.5 Double exponential smoothing forecasts 192
13.6 Effects of sudden demand changes 193
13.7 Creating the base series 196
13.8 Forecasting with base series 197
13.9 Baysian approach 199
13.10 Forecasting models 199
14.1 Contrasting independent and dependent demand 202
14.2 Inventory and MRP 202
14.3 Process sequence 204
14.4 Contrasting dependent and independent demand approaches 204
14.5 MRPII structure 205
14.6 Basic MRP calculation 207
14.7 Bills of materials 208
14.8 Simple master schedule layout 211
14.9 Master schedule supply planning 212
14.10 Control of supply and demand 214
14.11 Available to promise 215
15.1 Inventory location options 223
15.2 The 9-box model 224
15.3 9-box structure for supply 225
15.4 Availability structure 225
15.5 Central and local management of inventory 226
15.6 Supply inventory options 228
15.7 DRP stock and forecast demand 232
15.8 Inventory available for fair shares 235
15.9 Fair shares allocation 235
Preface
You are looking at this book because of an interest in solving a problem. This book
is written for the practitioner who wants to improve inventory performance and
the student or new inventory manager who needs to understand which techniques
to choose to get the best results, and how they work.
In the distant past, when I ascended from being a scientist to the heights of
inventory analyst, I was surprised at the lack of helpful written work on the subject.
Although there are now many books, manuals and papers, there is not much writ-
ten on the very important subject of how to do inventory control faced with the
real inventory, customers and an assortment of inventory records and styles of
suppliers. We have ‘big data’, but what do we do with it? Better control of stock
(inventory) can give major benefits to the profitability of all companies in a supply
chain. My experience in using the simple ideas in this book has been that avail-
ability and profits have improved whilst inventory value has been reduced (in
several cases by millions of pounds sterling).
This book explains how to use good techniques in the right combination to
improve performance in practice. It distils techniques which have been tried out
and proven to work. It is a result of working continuously on inventory control
with a large number of companies over many years.
This book will prove invaluable to:
• students of supply chains, operations management or even purchasing;

• inventory managers who need to enhance their status by improving results.
Many people do not understand that professional expertise is required to control

inventory. This book is a driver’s manual for inventory controllers. It will cover:
the working of the engine (how inventory control techniques work); how to use the
controls (what the techniques do and how to manage them); and how to get the best
out of the vehicle (how to optimise inventory). Understand the text and it will
show the way to guaranteed improved inventory control, reduced stock levels and
higher availability. The software will not do it properly unless an expert is in the
driving seat.
This book is the result of contributions from the many thousands of people with
whom I have worked, lectured, debated and implemented the techniques. Thanks
Preface xiii
are due to those companies who have let me loose on their inventory, carrying out
consultancy projects which have extended my knowledge and helped to hone the
concepts.
I lecture for the University of Warwick and CILT (previously IOM and BPICS),
so my students have been able to question and develop the ideas, especially at a
time of change where communications are excellent and many supply chains are
becoming closely integrated.
This third evolution of the book has been a two handed operation. The concepts
have been blended into a logical text, and I am indebted to Elaine Duckworth, who
has shared the task of developing a useful and readable handbook.
Dr Tony Wild
Introduction
• Why we need professional inventory management.

• Structure of the book.
• Focus on developing theory that works in practice.
Everyone is an inventory controller, at home and at work. We all keep food,

clothes, domestic items, images, tools and many other goods. We also have
shortages and emergency purchases. Some people regularly have to throw out the
contents of the ’fridge because it has been there a while and changed in character:
so, inventory control is a natural occupation which everyone does, some more
successfully than others.
There are many other activities which people do, such as sport, music, cookery
and medicine, and for these it is generally accepted that a level of technical exper-
tise is needed. In music and cooking there are people who become highly skilled
through their natural talents. Success appears to come from a mixture of talent,
determination, practice and knowledge. In medicine, particularly in surgery, we
are more likely to rely on those who have studied the subject and have a deep
knowledge of both theory and practice. Lawyers cannot function without
knowledge; often a common-sense approach does not work for them.
As with any profession, inventory control (or the subject, Professional Inven-
tory Management) has a body of techniques and knowledge which differentiates
the professional from the DIY enthusiast. This expertise is the result of 100 years
of development and refining. There are still colleagues who believe that inventory
can be managed using a clear brain, good communication and common sense. It is
unlikely that they would be happy with the same competency in their doctor or
financial advisor. Many executives are unaware that inventory expertise even
exists and so miss out on operating excellence.
Inventory management is a well-developed science, not simple

common sense.
Introduction xv
This book is for inventory control practitioners and students. The techniques
described here have enabled many people to manage their inventory so that their
customers are happier, and so are their accountants. The reduction of inventory
value, the avoidance of unnecessary work and the improvement of customer
service can be accomplished at the same time through simple application of the
combination of techniques discussed in the following chapters. The effect could
be to halve stocks and improve service at the same time.
The techniques described are the concepts that inventory controllers and
students should have at their fingertips. The more complex theoretical approaches
can be left to those developing sophisticated systems, or those with excellent
inventory control (and over the years I have not met any of these!). Inventory
practitioners should be able to use this book to understand the best approaches and
then to apply them to their own circumstances. Direct application of the methods
through the latest software, using spreadsheets or basic controls is most successful:
usually modifications result in less effective outcomes.
This book covers the academic requirements for many supply chain and operations
management courses, including inventory and supply chain theory and the practical
application of the techniques. The structure of the book is to begin with inventory
structure through target setting in Chapters 1 and 2, and how to structure inventory in
Chapter 3. Chapter 3 shows how a big impact on inventory levels can be made easily.
However, in practice, businesses end up with excess stock, so Chapter 4 shows how
to get rid of it. These chapters enable operators to gain kudos using the techniques;
however, the achievement must be focused by the correct management targets within
the appropriate management structure, which is the topic for Chapter 5.
Within this framework, lean supply could feature as the best way to organise
(Chapter 6) or, failing that, the classic approach to inventory levels is best, so
Chapters 7 and 8 are the elixir of item-level stock control – the secret recipe for
success. Chapter 7 concentrates on how to keep out of trouble – a rational approach
to the risks of stock-outs – and Chapter 8 shows how to set stock levels. Repetitive
supply requires close supplier relationships and delivery structures, which are
covered in Chapters 9 and 10.
An aspect of inventory management that is usually carried out poorly is fore-
casting. Forecasting methods are relatively reliable, but not well used in practice.
This is an opportunity, introduced in Chapter 11 and detailed in Chapters 12 and 13.
The techniques are not too mathematically complex, but have been proven to give
good results in practice. Ways to ensure that forecasts and safety stocks produce
sensible results are also described.
Manufacturers use items in balanced sets, and this dependent demand can be
useful in many other situations by creating a master for associated demands and
sticking to it. Chapter 14 describes requirements planning and how to use it.
The cumulative effect of the techniques used throughout the supply chain will
have beneficial effects, but co-ordinating activities along the supply chain has
much more potential. In Chapter 15 the risks are identified and the discussion
there shows how to achieve the potential benefits, built upon the expertise
developed throughout the book.
xvi Introduction
For people who want to ensure their understanding of the concepts and
quantitative solutions, there is a selection of questions (with answers) to test this
at the back of this book.
As the information in this book has been gleaned from many discussions and
conferences, interpreted and used successfully by the author, and from very little
literature, there is a dearth of direct links to references of previous work. Inventory
expertise is based on blending the excellence and understanding of colleagues,
practitioners and academics.
Readers outside the UK may find that a little translation is required. For instance,
in America the term ‘inventory management’ is applied to the whole subject,
whereas in the UK it is normally known as ‘stock control’. The terms ‘inventory’
and ‘stock’ are both used in the text to mean the same thing (in the UK inventory
has a financial flavour and stock an operational context). It is hoped that this and
other nuances of translation do not cloud the understanding of this technical and
fascinating subject, whatever it is called.
If readers would like presentation material or video, then it is available on the
companion website: www.routledge.com/cw/Wild. A complete set of PowerPoint
slides of the figures and some additional material is also available to registered
readers at the same website.
The author is available for discussion by email at dawsonberkeley@btinternet.
com.
1 The basis of inventory control
• Conflicts which beset inventory controllers.

• What inventory is there to achieve.
• How good inventory management will improve profitability.
• The reasons for the current stockholding.
1.A The role of inventory management

The success of a venture depends on its ability to provide services to customers or
users and remain financially viable. For an organisation that supplies goods to its
customers, the major activity is to have suitable products available at an accept-
able price within a reasonable timescale. Many parts of a business are involved in
setting up this situation. Initially it is the marketing and design departments, then
purchasing and, in some cases, manufacture is involved. For an item which is
already in the marketplace, the main activity is providing a continuity of supply
for customers.
Inventory control is the activity that organises the availability of items to
customers. It co-ordinates the purchasing, manufacturing and distribution func-
tions to meet marketing needs. This role includes the supply of current sales items,
new products, consumables, spare parts, obsolescent items and all other supplies.
Inventory enables a company to support the customer service, logistic or man-
ufacturing activities in situations where purchase or manufacture of the items is
not able to satisfy the demand. This could arise either because of the speed of
purchasing, or because manufacturing is too protracted, or because the quantities
demanded cannot be provided without stocks (i.e. the demand rate exceeds the
maximum supply rate).
Stock control exists at a crossroads in the operations of a company (Figure 1.1).
Many activities depend upon having the correct level of stock, but the definition
of the term ‘correct level’ varies depending upon who defines it. Stock control is
definitely a balancing act between the conflicting requirements of the company,
and the prime reason for the development of inventory management is to resolve
Customer
service
Operating Low
efficiency inventory
Purchase
price reduction
Figure 1.1 Which direction?
this conflict in the best interests of the business. A conventional supply organisa-
tion will have many departments, including sales, purchasing, finance, quality
assurance, contracts and general administration. In some cases there will also be
manufacturing, distribution or support services, or a variety of industry-specific
activities. Each of these has a biased view of the ‘correct level’.
The strategic view of inventory management is that it exists for three reasons:
1 To cover up the mismatch between supply and demand processes.

2 To minimise the risk of failure to supply.
3 To minimise the overall costs in the supply chain.
Inventory separates demand from supply, so that users are not affected by the
limitations in the supply chain.
Inventory is the fluid that lubricates the wheels of the supply chain.
The aim is to provide the most beneficial solution to the end users and the
providers within the supply chain. This could include short-term profit, long-
term growth or other advantages, such as market retention. Where the strategy
can include co-ordination along the supply chain, stockholding can be struc-
tured to provide maximum availability and minimum inventory, consistent with
minimum risk.
The inventory manager needs to understand their colleagues’ differing views of
the requirements for inventory.
The basis of inventory control 3
Sales consider that good stock control enables the company to have available
any item which will meet immediate sales for as large a quantity as demanded:
this requires large stocks. For service companies where parts are required, the
control of stock at the customer interface is traditionally left with the person
carrying out the service, which normally causes stock-outs and panic supply.
For efficient distribution, shipments should be in full loads, which can lead to high
stock levels. A compromise has to be reached.
Inventory control keeps balancing conflicting requirements.
Purchasing consider that inventory is a buffer enabling goods to be purchased

ahead so that optimum prices can be obtained. Buying items in bulk often reduces
the purchase price and it also reduces the workload within the purchasing
department. The stores are a means of keeping bulk purchase items after buying
advantageously.
Finance departments have problems with stock because it consumes vast
amounts of working capital and upsets the cash flow. One benefit of stock from a
financial standpoint is that provisions can be made in case the stock turns out to be
unsaleable, and this value can be adjusted to modify the profit figure in times of
good or bad financial results. However, the existence of these provisions in the
first place is detrimental to the company’s finances.
Reliability and consistency are the keys to effective inventory management.
Quality management often causes delays in processing materials and increases

in recording, and therefore greater inventory. Applying improved quality man-
agement systems can avoid this problem by ensuring consistent operation of
supply and processes, rather than the products. Businesses should review whether
any quality checking can be eliminated by co-ordinating with suppliers.
Logistics view inventory as an advantage in maintaining full use of distribution
so that load size can be optimised and distribution costs reduced.
General management see stock control, rightly, as a source of information.
Some management consider that they should be able to use stock control to give
an immediate supply of information, statistics and forecasts. This can result
in large amounts of unstructured work collecting, analysing and providing
information.
1.A.1 Manufacturing
The traditional view of manufacturing companies has been that large batches
reduce direct production costs. Manufacturing management tend to aim more for
plant and labour efficiency and allow high stocks in order to avoid the disruptions
caused by shortages, breakdowns and changing customer demand.
Good stock controllers (or materials management) will keep the stock down
as long as they are responsible for stock levels of everything, including raw materi-
als, finished and semi-finished stocks, consumables, tools, work on the shop floor.
Do not confuse high stock with good availability.
The focus is now on producing variety, ideally at the same rate as the end user
is consuming the items. For manufacturing, batch size is often linked to set-up
(changeover) time, which is mistakenly accounted for as a direct cost instead of an
overhead. Lean supply has led the way to providing low inventory with low
operating costs. Continuous improvement in quality, reliability, delivery frequency
and processes reduces inventory and enhances customer service. Gradual changes
led by operational people (known as the kaizen approach) help to maintain
competitiveness; for example, a change from planning using MRP (material
requirements planning; see Chapter 14) to using kanban (see Chapter 6).
The responsibility for maintaining the correct balance is normally left to inventory
management. (The actual job title could be controller or manager of materials,
inventory or supply chain, or stock planner, or even considered within purchasing,
warehouse or logistics in small organisations.) Companies should be careful that
inventory is not left in a gap between supply and sales. The proper process is:
Forecast Sales ⇒ Determine Appropriate Inventory Supply ⇒ Replenish Inventory
Demand for the products varies as a result of changes in market and financial
forces. The amount and type of stock control vary in tune with this. Stock control
is a dynamic activity, and the successful inventory manager has to ensure that the
balance is kept right. This requires both communications skills and professional
inventory techniques. Operating methods should be revised continuously to reflect
the changes: systems should be altered to suit new situations and operating
policies.
1.B Objectives for inventory control

Initially, the reason for stockholding has to be determined. It could be to achieve
good customer service, cheap purchase, bulk discount, security or other corporate
objectives. The stockholding strategy depends directly on the company objectives
and policy, and identifies:
• the service that it intends to provide to customers;

• what investment to make in stock;
• how orders are placed on suppliers;
• the principles for controlling throughput of items for the business.
This is the first ground rule of stock control – to have a policy worked out
by top management to guide the operation. If stock control is done profession-
ally, more effective targets can be set. This enables businesses to manage against
more challenging KPIs (key performance indicators). The company policy can
then be interpreted into detailed plans and focus the balance in stockholding and
operation control.
The overall strategy for inventory management, like all other activities in the
company, is to contribute to the welfare of the whole organisation. The inventory
and logistic operation must aim to contribute to profit by servicing the marketing
and financial needs of the company. Stock control is central to achieving this. The
aim is not to make all items available at all times, as this may well be detrimental
to the finances of the company. The normal role for stock control is to meet the
required demand at a minimum cost.
Our aim towards long-term profitability has then to be translated into opera-
tional and financial targets that can be applied to our daily operations. The purpose
of the inventory control function in supporting the business activities is to optimise
the three targets:
• customer service;
• inventory costs;
• operating costs.
The most profitable policy is not to optimise one of these at the expense of the
others. The inventory controller has to make value judgements:
• If profit is lacking, the company goes out of business in the short term.
• If customer service is poor, then the customers disappear and the company
goes out of business in the longer term.
Balancing the financial and marketing aspects is the answer – the stock controller
has a fine judgement to make.
The first target, customer service, can be considered in several ways, depend-
ing on the type of demand (discussed in Chapter 2). In a general stores environ-
ment, the service will normally be taken as ‘availability ex-stock’, whereas in a
supply to customer specification, the service expected would be delivery on time
against customer requested date.
The second target, inventory costs, requires a minimum of cash to be tied up
in stock. As the target is in terms of total value, the focus has to be on items where
the turnover value is high. However, there are so many low usage value items
that there need to be limits because of the cumulative value and the risk of
obsolescence.
The motivation for maintaining low inventory may also arise from minimising
other critical resources. The physical size of bulky items has to be considered
when optimising warehouse space, distribution loads and container shipments.
The third target, avoiding operating cost, is controlled by the way inventory
processes are organised. The prime operating costs are warehouse operations,
inventory control, distribution logistics, purchasing and the associated services.
Improvements in the supply chain operations through consistency, reliability,
demand smoothing and collaboration have provided major benefits.
Inventory management is simple:

Keep customers happy, stock value low and reduce the operating costs.
Optimising the balance of these three objectives is the focus for inventory man-
agement: the better the balance, the greater the profits provided (as illustrated in
section 1.C, profit through inventory management). The development of supply
chain co-operation, lean thinking and communications technology has led to
major opportunities to improve. Companies who tighten their control faster than
the average will flourish, but those that do not keep changing methods fast enough
will gradually dwindle. The way to improve is relatively simple, and is outlined in
the remainder of this book. The trick of the good inventory controller is to meet
the objectives simultaneously, not one at a time, and of course ‘the better the
control, the smaller the cost, the lower the stock levels and the better the customer
service’.
In reality:
More inventory = worse delivery on time.
One of the dichotomies of inventory control is that, at item level, the more stock
the better the availability (this is discussed quantitatively in Chapter 7, Safety
stocks). However, for the whole inventory, experience has shown that businesses
with the highest stock are often those which have the worst availability. These
observations are not in conflict if the causes of this are considered. Stock-outs
result from holding too little stock for the offending lines, because the forecasts,
monitoring or controls are inadequate. High stock levels arise because too much
stock has been purchased, through bad forecasting, monitoring or controls. High
stock and poor availability are caused simultaneously as a result of poor control.
The problem rests with the inventory controller, and the solution is in improved
techniques. Using the techniques discussed in this book, the reader can reduce
stockholding by a quarter, even where the control is already reasonable.
1.C Profit through inventory management
There is obviously a link between financing stock and profits. This was the origi-
nal reason for the author taking up this profession.
The profits of a business reflect its long-term success. The bigger the company,
the larger are the expected profits. This profitability is most commonly calculated as:
Profit
Return on Sales =
Annual Sales
and
Profit
Return on Assets =
Assets Employed
The assets include both:
• fixed assets, forming the permanent capital of the business (plant, buildings,
land, systems, infrastructure, etc.);
• variable assets that change continuously (all types of stock and debtors).
and the effect is clearly illustrated by the following example.
1.C.1 An example of the effect of stock on profitability

V.R. Slack Inc. set up a subsidiary company in the UK seven years ago, and
shipped across large inventories in order to ensure customer satisfaction. Success
caused the business to expand and the inventory increased too. They gradually
increased the turnover to £5 million and inventory to £2.5 million. As a conse-
quence, they had to rent warehouse space and £2 million of stock is now in this
warehouse.
Four years ago, a small company (M. Tight Ltd) was set up in opposition by
some ex-employees of V.R. Slack. As they were undercapitalised, they had to buy
and sell quickly to maintain cash flow. As a consequence, they concentrated on
understanding customer demand, and although their turnover reached £5 million,
their inventory was only £0.5 million.
Storage costs and inventory consume cash.
V.R. Slack have had to finance this extra £2 million inventory. They were lent
this by their corporate headquarters, who expect a 15% p.a. return on investment,
so V.R. Slack pay £300,000 p.a. interest. They also have the costs of controlling,
Cost element Cost (£ p.a.)
Stores: extra area rent, heating, etc. 500

Stores, equipment; trucks and racking
depreciation 1,000
maintenance 1,000
Stock obsolescence 5,000
Storekeeper’s wages and employment costs 14,000
Annual stocktaking 500
Control systems and computer time 1,000
Total 23,000
Plus financing charge £2 million @ 15% 300,000
Total extra cost £ 323,000
Figure 1.2 Cost element.
holding, transferring and organising the stock stored in the rented warehouse.
These additional warehousing costs are shown in Figure 1.2.
Now examine the Profit and Loss Account for the two companies. Two years ago
the difference between sales revenue and costs was £250,000 for each company.
M. Tight have a return on turnover of 5%, making £250,000 profit on the
£5 million sales. V.R. Slack had extra warehousing costs of £323,000 and thus
made a loss of £73,000. The only difference is caused by the stockholding.
The next year the businesses were able to double the profits (£500,000) and the
inventory levels and turnover stayed the same. So, M. Tight made a ‘return on
sales’ of 10%. V.R. Slack’s extra inventory was still £2 million, so the profit was
£500,000 – £323,000 = £177,000 because of their extra stockholding costs. So,
V.R. Slack only had a return on sales of
177, 000
= 3 12 %
5, 000, 000
This shows how important it is to the profit of a company that stockholding costs
are minimised.
However, the most important issue is return on capital, where the situation is
even more dramatic. Both companies have plant and buildings worth £2.5 million
(fixed assets). In addition, they have stock (£2.5 million for V.R. Slack and
£0.5 million for M. Tight). In their industry, everyone pays directly so there are no
creditors or debtors. Consequently, total assets are £5 million for V.R. Slack and
£3 million for M. Tight. The results for the second year are therefore:
500, 000
The return on assets for M. Tight is = 16.67%
3, 000, 000
For V.R. Slack, the return on assets is therefore:
177, 000
= 3.54%
5, 000, 000
V.R. Slack waste their profit by not managing inventory properly, but they do
not realise that it can be done better. Most companies are like this.
Hold a minimum of the right items to be profitable.
M. Tight has cash available for investment, wages and better working conditions,
and for shares and pension schemes as well as a return for investors in their company.
V.R. Slack has little cash to spare other than to pay off the debt. Imagine the
situation: poor profitability leading to disillusionment, poor working conditions
and fears for the future.
1.D Reasons for the current stock

Inventory exists because supply does not match demand. It is easy to assess the
effectiveness of inventory control in a company. Just visit the warehouse.
Your current stock shows what is going wrong.
The stock there is a result of the effectiveness of stock control, but the stock
there is probably not ideal! Instead of discussing the reasons and blame for each
item (and there are some interesting causes for stockholdings), the inventory can
be classified according to the root causes of the stock. An examination of the fac-
tors that give rise to the stock will identify the strategy to reduce the levels. There
are nine major underlying causes, as follows.
1.D.1 Supply quantity

What is the ideal batch size? Many items are bought in bulk, but issued and sold
in smaller quantities. At any time, there is a balance of receipts still in stock – a
half empty box or pallet. There is always a challenge to reduce supply quantities
without increasing costs and move towards inventory flow along the supply chain.
Meanwhile there is some ‘batch quantity stock’ that contributes to inventory value.
As transport costs are often significant, inventory controllers have to be careful
to transfer a wide variety of items in each shipment, or change shipment method
in order to minimise inventory and logistic cost.
Frequent, small deliveries keep stock low.
Where items are being manufactured, the rate of production at one stage of the
process may be greater than at other stages. The ideal solution to this would be
kanban (see just-in-time supply, section 6.B.3), but often there are stocks of items
manufactured slowly waiting to fill a process load. Alternatively, a bulk process,
or a relatively fast process producing items at a fast rate, will create stocks which
are subsequently used up at a much slower rate.
1.D.2 Safety stock

In many situations customers do not provide information of their demands far
enough in advance. To compensate for this problem there are two tactics:
• The first is to organise the customer to give more forewarning. Is this impossible?
Suppliers of specialist, make-to-order or highly sought after products may make
their customers wait until the product is ready. In more competitive situations,
however, the customer has more power and may not be prepared to wait. This puts
the pressure on suppliers to reduce the lead times and to forecast demand better.
With better knowledge of the customers’ demand, the unexpected peaks would no
longer be a surprise and stocks could be varied to cover these occasions.
Let the source of the risk hold the inventory.
• The second option is to hold sufficient extra stock to cope with unexpected or
excessive demand. Safety stock (or buffer stock, as it has been called) is there-
fore added to cover our inability to predict demand. The inventory manager has
an investment choice. Either invest in inventory or invest in information. Those
who can predict customer requirements accurately do not need safety stock.
A secondary role for safety stock is to compensate for failure on the supply
side – non-delivery of purchased goods, information failures, mechanical
breakdown or even industrial action. ‘You can’t be too careful!’ Generally, the
approach to these situations is to monitor the situation continually and make
gradual improvements (see Effect of supply lead time, section 7.C.2).
1.D.3 Market change

Customers change their minds, contracts are lost, markets vary. These all cause
excess or even obsolete stocks, but this should not be a significant value. The
changes that cause a build-up of inventory are either step changes or gradual
changes. A gradual change is the most common and leads to lots of minor excesses
of stock. The cause is normally a forecasting method that is not reactive enough,
and the simple cure is often to use a more advanced forecasting method. Methods
discussed in Chapter 13, used in the right way, avoid this situation.
Detect trends and react automatically.
A sudden reduction in demand can only be anticipated through understanding

and information. Input from sales personnel is important in avoiding excesses as
well as stock-outs. Sales personnel can warn of impending major changes, even
though they are poor at identifying the size and timing of changes. This is the best
way to avoid unsaleable items remaining after demand is satisfied.
1.D.4 Obsolescence
In addition to excess stock caused by customers, a significant amount of obsolete
stock can be caused within the company itself. Consider several situations:
1 The marketing department puts items on promotion without informing the

inventory controller. The demand increases as the promotion starts, inventory
control sees demand escalate and stocks run out. They therefore order some
more. The promotion ends just as the new stock arrives. The company now
has three years’ worth of stock cover at the normal usage rate, and major
customers have already bought extra during the promotion.
2 The design department has replaced the old version of a product with an
improved one, the sales team were extremely pleased with the new product
and started selling it immediately. However, there was a large amount of the
old product left in stock that was not sold.
3 The marketing department has just launched the new season’s catalogue.
It contains some new items and some old ones have been deleted. There are
still good stocks of some of the deleted ones.
4 The sales director has bought a job lot of items, which are going to be sold as
a new line.
All these are common occurrences in some businesses. They all cause
self-inflicted obsolete stocks and can be avoided by improving communications
between departments. Stock managers have to organise this, since the other
departments rarely understand the challenges of balancing inventory or the
concept of lead time (see section 4.C, Obsolete and excess stock).
1.D.5 Poorly defined responsibility

When buyers procure to optimise purchasing costs and their operations, and sales
sell to suit customers, in the middle are the stores, accumulating stock. Stock
levels are unbalanced and control becomes powerless, simply recording and
storing. Inventory management should have the responsibility for assessing the
demand pattern, determining the appropriate stock levels to support this and then
instructing purchasing what and when to supply (as discussed for manufacturing
in section 1.A).
1.D.6 Planned inventories

Seasonal demand, cyclic demand or demand to meet a known peak may be
supported most economically by stock building ahead. Some stores’ stock may
be there, building up in advance of a demand event or stored to cover a definite
requirement. The feature of these is that they cover situations where the demand
events are certain and in excess of the supply capabilities. For example, in indus-
tries where the supplier has an annual holiday shutdown, stock has to be provided
to cover customer demand over this period, since it is not possible to acquire extra
quantity at the last minute to cover the shutdown. The stock is therefore gradually
increased over several weeks beforehand to support the ongoing demand. This
effect can be complicated by different stages in a supply chain taking holidays at
different times.
1.D.7 Layout and location of stores

The effects of stock location must be analysed in the context of the whole business
or the whole supply chain. If stock is located a long way from where it is required,
then the users tend to keep their own stock. In fact they might hoard and take more
than they need, which causes the main stock to run out for everyone else. This lack
of co-operation and control can result in satellite stores being set up and excess
stock being held.
A worse situation results if different departments order the same items
independently. In this case, there are totally independent stock systems, usually
controlled by managers who are amateur stock controllers, and, as a consequence,
stocks are excessive. Fortunately, this normally happens for low value consumable
items (e.g. office supplies).
Where a company has more than one store (such as in distributed warehousing),
a similar range of stock items are held in each, resulting in duplication of stock.
If there are slow-moving items in one store (try and find a store without slow-
moving items!), then the same item can be found in another store – still moving
slowly – and total stock across all the stores can exceed the annual demand. This
is a natural result of distributed stock. The safety stock required will increase as
the square root of the number of distributed stores, so managers had better think
carefully about setting up local stock points. The increase of stock (including the
square root) is based on the assumption that the local stores are part of an inte-
grated, well controlled inventory system. If this is not the case in practice, then the
total stock can go through the roof (literally).
The study of inventory in a supply chain usually reveals that there is stock held
by suppliers, and the same stock is held by the customers because they do not trust
the supply. Modern supply chains organise the responsibility, so that this duplica-
tion is avoided. A principle of supply chain management is:
Hold inventory as far back in the supply chain as possible while maintaining
customer service.
1.D.8 Company strategy

Some inventory controllers do not have complete authority to maximise avail-
ability and minimise costs. This may be because top management have imposed
constraints, often extra stockholding, because they have little confidence that the
stock system provides adequate customer service consistently. Sometimes sales
management will even agree to provide consignment stock (see section 15.D)
or make a commitment to keeping an agreed quantity of inventory, instead of the
sensible option, which is ‘agreeing an availability level’.
Top management should set strategy and targets, and provide support so they
are achieved. They should employ professional, competent inventory controllers
to organise the detail and methods through which these are achieved. It should be
extremely rare for a good executive to get involved in specific issues.
1.D.9 Systems and control

The time taken to process information can substantially affect the level of
stockholding. Consider an issue from a store that makes the stock level fall below
the ordering trigger level. How long does it take before the supplier gets to know
that a new delivery is required? On a conventional order processing system, the
procedure is:
• Enter issue from stock on system (any delay here?).

• Allow purchasers to interrogate system for items that need ordering (next
morning?).
• Confirm how many of the items to order (rest of the day?).
• Get authorisation for purchase (when the director is next available!).
• Transfer information to supplier (immediately by web?).
These processes could take a long time, especially if many different lines are
batched up into a large order. If the supplier is a local stockist, then the item could
arrive the same day, but the total supply lead time could be one week because the
stock is only reviewed once a week, so the stock level would have to be corre-
spondingly large to compensate for this.
For efficiency, use real-time procedures.

The solution is to use technology properly, avoid paperwork and link with
suppliers and customers by integrating systems, understanding their needs and
using social media so that trends can be spotted.
1.D.10 Analysis and reduction

If the stock is analysed according to the nine headings above, the causes of stock
excess become obvious. The results will show where the stock balance is not right
to meet marketing and financial targets in the most effective way. By establishing
the major reasons for this, a plan can be devised and action taken to ensure the
lowest stockholding levels necessary to meet company targets.
Case study
An independent non-ferrous metal distributor in Leigh, Lancashire,
found that their stock value had been growing relative to turnover. The
total inventory represented about six months of demand. They were
looking to improve the situation and called us in as inventory special-
ists. An analysis of the stockholding had already been done, and iden-
tified that some 20% of the stock value was slow-moving and
non-moving stock. They were working hard to reduce the slow movers.
Our approach was to carry out the analysis into the nine causes
described. From this we found that the major contributors to inventory
value were:
Purchase order quantity stock 30%

Safety stock 25%
Obsolete stock 20%
Total of other six causes 25%
(The assessment of demand was via customer schedules and quite

accurate.)
The advice given to the company was to concentrate initially on
purchase order quantity stock and safety stock. There were fewer lines
to work with and reducing the slow-moving stock is laborious and
long-winded.
The plan was to get more frequent deliveries of all the high spend, faster-
moving items. Since the key suppliers delivered weekly, this was achieved
after a negotiation (resulting in no cost increase). This halved the delivery
quantity inventory value. We also carried out proper evaluation of the
safety stocks, which resulted in a reduction in that inventory by some

35%. Our input to the whole process was some four days’ work, and
the company spent an equivalent time talking to suppliers. At the same
time, some obsolete and slow-moving stock lines were selected for the
sales people to offload.
Of course, the stock reduction did not happen immediately – it took a
couple of months for the stock to go down – but it was not difficult to
achieve.
The techniques used were:
Order quantity reduction – Pareto Analysis and supply scheduling

Safety stock rationalisation – safety stock calculation
as described in later chapters for you to use.
Key points
• Consider the objectives and need for inventory.
• View inventory management as a core activity in ensuring the
success of the business.
• Target availability, inventory value and operating costs simultaneously.
• Create profits by avoiding non-essential inventory.
• Determine the causes of stock and use techniques to reduce it.
2 Customer service
• Identify the purpose of customer service.

• Define what availability the customer needs and measure it.
• Enter the demand and make sure the customer is happy.
• Analyse the components of administering customer service.
• Discuss the role of forecasting and availability.
Companies live or die through their overall service to customers. Customer service
is a complex subject of its own, but there are two main aspects to meeting customer
requirements, namely:
1 customer relations, and

2 availability of service or items.
2.A Customer relations

It requires interpersonal skills to ensure that customers have the correct level of
expectation of supply and that expectation is met. If they are happy with their
purchase, there is potential for repeat business and wider sales.
This aspect of customer service has gradually emerged because it is the differ-
entiating factor between many companies. The product can be similar from a wide
variety of suppliers; the vendor that is most successful is the one whose quality
customers like and trust.
Fix the customer, don’t just supply the item.
Customers’ perception of suppliers varies greatly from their actual perform-

ance. It does not matter whether suppliers are good or bad – what does matter is
what the customer thinks about them.
Customer service 17
Customers can look for a match between their own style of company and the
style of the supplier. If there is a match, then relations are likely to be good; if not,
then the co-ordination requires more work. The major factors affecting customer
relations are a result of market strategy projected through sales and sales order
processing. However, there has to be the same customer-oriented approach in stock
control, which, like the company, needs to maintain reliability and credibility.
Availability, the second aspect of customer service, is a key target for inventory
control. It is easily measured and can be achieved through the application of the
techniques described in this book.
Example
A supplier of fasteners has trouble with widely fluctuating demand levels
on some products. This is because the customers think ‘I need a box of
fixings for this job’ and order enough to do maybe two months’ work.
Consequently, the supplier gets hit by two months’ demand and then
nothing. This is acceptable if there are many customers, but for the
slower-moving fasteners, demand can be very erratic. By understanding
the supply problem, the customer can be encouraged to order enough
to start the job off with sufficient supplies, while the supplier, who under-
stands the user’s situation, can provide the range of fasteners but can
get away with much less stock. Of course, the situation must be care-
fully monitored and the balance must arrive before the initial stock has
been used up.
Large quantity orders can be filtered out using simple criteria (see
Chapter 11) so that real demand priorities can be fulfilled. Demand can
thus be managed through the order processing activity.
2.B Measuring availability

How good is our service to the customers? By measuring availability, a company
can assess whether performance is likely to be acceptable. However, there are
many ways to carry out this measurement.
2.B.1 Availability
The entire reason for stockholding is to have items available. Woe betide busi-
nesses that do not manage their service level.
For each item in stock, the risk of stock-out can be reduced by increasing the
stockholding. The larger the investment in inventory for an item, the better the
service. This is shown by the solid curve in Figure 2.1 for an individual stock item.
The curve shows that it is relatively inexpensive to give a reasonable level of
18 Customer service
Figure 2.1 Availability of a stock item.
availability, but the value of stock increases very rapidly when we try to achieve
high levels of service. Companies will improve availability by increasing investment
and moving up the solid curve. To go along the arrow toward better availability at
less cost means changing the method of operation and improving the forecasting,
as described in subsequent chapters.
100% service is not possible because it needs an infinite amount of stock!1
However, 99.9999% availability is possible, if very expensive, and no one can tell
the difference from 100%.
Availability: get the facts, then act.
There is obviously a trade-off between financial investment and the availability

of items from the shape of the curve. The best trade-off results from all items being
at the same level of availability, appropriate for the market being supplied.
The overall service to customers, or other users, is measured by success in
meeting demands (on time availability). It is calculated on a daily, weekly or
monthly basis by adding the service provided on all the individual items, taking
into account the issue frequencies for the total range of items. We can define avail-
ability as:
demand satisfied
availability =
total demand
This criterion should be applied to all areas of inventory to monitor how well the
investment in inventory is doing.
Customer service 19
When the whole range of stock items is considered, the minimum stock level results
from all the items being under control at their optimum service level. Basically, all
items should be at the same point on the curve in Figure 2.1. The only way to reduce
stock and maintain availability is to change the management processes. This should
give a higher curve (dotted), with the same availability at lower inventory. The
improvements aim is toward the top left corner of the graph – high service with low
stock across the whole of the inventory.
High stocks go with poor availability – they are caused by poor control.
2.B.2 Measurement of availability in practice

Within the definition of availability, there are several ways by which the service
level can be measured, depending upon the type of demand.
For companies supplying single items to individual users as a result of many
individual internet, telephone or mail orders, then the definition could be:
total number of items supplied

=
total number of items ordered
r
For a department which is sending large export orders, these are often covered
by a ‘letter of credit’ or the whole shipment must be sent in one crate. In this case,
the service level would be:
total number of complete orders supplied

availability =
total numberr of orders required
For businesses where the demand is not required to be satisfied immediately,

care must be taken to measure availability at the time required for despatch, not
the time of receiving the order. This gives the inventory controller the opportunity
to fulfil the demand from elsewhere rather than from stock.
Care must also be taken as to the definition of ‘on time’ delivery. This is the time
specified by the customer originally without any renegotiation. Delivery should
really be measured as the time the customer receives the stock rather than when it
left the stores. ‘On time’ is when the customer needs the stock. Where there is a
history of poor delivery, the customers are quite likely to request a delivery date
well in advance of their real need, in order to have a chance of having the items
when they need them.
Measure availability in line with typical customer needs.

20 Customer service
The type of customer also affects the measurement of availability. For example,
if a customer requires ten but the stock is only eight, what customer service can be
provided? Is the answer:
• 0%, because the demand was not completed? (non-stock/ manufacturing answer)
• 80%, because eight out of ten can be supplied? (normal answer)
• 100%, because the balance can be provided before the customer needs them?
(for stockholding customers)
Different situations have different priorities. To have an OTIF (On Time In Full)
is often quoted by purchasers. This is theoretically impossible unless a maximum
demand rate can be agreed with the customers. Figure 2.2 shows the options.
There is normally a choice between full load delayed or part ship on time. Where
OTIF is essential (e.g. automotive component suppliers), the demand rate is level
and is forecast quite accurately, so OTIF can be achieved without high inventories.
For companies to measure availability, the service must be defined in one of the
above ways – the one which suits most customers.
In Figure 2.2, only two items are being considered. The demand for item
A12 was five orders, totalling 1,000 items, of which 990 were sent on time
and ten items were not available for one order. This means that 80% of the
orders (and 99% of the items) were sent on time. The second item is a slow
mover and there was only one order (for ten items) and only five were des-
patched on time, giving a 50% service of parts (and complete failure in
terms of complete orders).
How good is the service? Out of 1,010 items requested only 15 were
short, so the number of items despatched was 98.5% of the total. However,
in terms of item order fill, deliveries of item A12 were 99% full but B25 was
only 50% full, giving an average per line of 74.5%. If the customer is looking
for complete orders, then out of the total of six orders two were short, cor-
responding to a 66.7% success rate. As an order fill per line, this was 80%
for A12 and 0 for B25, so an average was 40%. This illustrates the different
assessment of levels of service that can often occur between customers and
suppliers. In this example, the supplier is claiming 98.5% delivery on time,
while the customer is measuring it at 40%!
Part no. Demand Shortage Service level (%) No. of orders Shortage Service level (%)
A12 1,000 10 99.0% 5 1 80.0%
B25 10 5 50.0% 1 1 0.0%
Average 74.5% 40.0%
Sum 1,010 15 98.5% 6 2 66.7%
Figure 2.2 Alternative measurements of availability.

Customer service 21
It is important to agree a common goal between suppliers and customers, and
to establish what the customers need in terms of availability. The appropriate
measure can then be used for monitoring.
2.B.3 Availability policy

Customer service can be structured or focused on a particular group of customers
or market sector using alternative availability policies. Options which could be
used are:
• the same availability across all the products;

• minimising the total cost of service;
• concentrating on the most valuable customers;
• enhancing the service on the most sensitive products;
• greatest availability on the most profitable products; or
• better service on the major turnover items, reduced service on slow movers.
These alternatives will provide different availabilities for different products, with
a management decision setting availability levels.
Whichever alternative is chosen, proper control of stock requires the inventory
manager to regularly monitor availability.
A better measure takes account of ‘lateness’.
Businesses also have to decide whether to measure ‘on time’ by the hour, day or
week. Companies mainly choose to measure by the nearest day, except where their
systems work weekly.
2.B.4 Back order measurement

The measurement of availability discussed in section 2.B.2 should always be used,
but a further measurement is necessary for sophisticated stock control.
Once an item has missed its delivery date, it is out of the calculation. This intro-
duces a further factor into the equation: the priority given to satisfying back orders.
Consider a case where there are two demands for an item: one which is past the
due date, and one which can be delivered on time (Figure 2.3). The availability
target calculation requires us to fulfil the current order, not the old one, since the
current one improves the availability score and the other has already failed. In
practice, there is normally a need to give precedence to customers whose delivery
is late. Therefore, a more rigorous measurement would include assessment of the
late orders and their degree of lateness.
A simple example of this approach is developed in Figure 2.4. It shows the
despatch position for two companies, A and B. Which company is providing the
Current
In Full
Performance
On Time
100%
Figure 2.3 On time or in full (OTIF).
Analysis Company A Company B

Delivered on time 10 30
1 week late 60 15
1 month late 20 40
3 months late 10 15
Total despatches 100 100
Figure 2.4 Comparison of despatches for companies A and B.

Items despatched in week 37: Analysis of lateness on orders despatched.
70
Company A
60 Company B
50
Percentage
40
30
20
10
0
Delivered 1 week late 1 month late 3 months late
on time
Level of fill
Figure 2.5 Despatch performance for companies A and B.

Customer service 23
better service? The graph prepared for management in Figure 2.5 illustrates the
situation well. The availability ‘level of fill’ (LOF) approach makes company B
the better one because it gives 30% LOF against 10% LOF for A. However, there
is a better chance of getting the items quickly from company A. Perhaps A is better
after all.
A measure of customer service that takes into account the overall perfor-
mance, not just the ‘ex-stock’ part, can be developed using the weighted average.
In Figure 2.6 the arrears, or ‘back orders’, have been analysed by categories of
lateness. This can be used for ex-stock and due date situations, and is illustrated
in Figure 2.7.
The number of items in arrears is the simplest measure, and the aim is to focus
the control on the important items. Hence a suitable measurement process is:
1 Record the number of late orders for each day (or week) late.
2 Multiply the number of late orders by the days (or weeks) late for each day
(or week).
3 Add up the total of these ‘days (or weeks) of lateness’ (see Figure 2.6).
4 Make the improvement of this figure a performance target.
Time overdue No. of items Weighted average

T (weeks) (N items) (T × N)
1 8 8
2 6 12
3 4 12
4 2 8
5 1 5
Total 21 45
Figure 2.6 Distribution of arrears.
14
Number and Weighted Average
Weighted
12 Average (T x N)
10 No. of Items
(N items)
8
0
1 2 3 4 5
Figures 2.7 Managing arrears.

24 Customer service
By multiplying the number of orders outstanding by the weeks of lateness, a
back order week figure has been calculated. This gives an overall ‘arrear weeks’
figure (45 in Figure 2.6). Inventory controllers have a better service monitor using
‘arrear weeks’. By continuously improving this, and avoiding making excuses,
customers will be placated. The figure shows the normal customer attitude that the
later the order (past the promised date) the less they like it. The inventory controller
has then to balance hard work getting very late orders despatched against ones that
have just missed ‘on time’ delivery. Obviously, the best practice is to ensure that the
bulk of demands are satisfied on time so that they do not get into this statistic at all!
Measurement in terms of value is not usually recommended since the aim is to
provide customer service. Routine assessment of arrears should be carried out in
a similar manner to the monitoring of availability.
Verify required dates before calculating arrears.
In practice, the order backlog viewed on the system may not be all true demand.
It often consists of:
• Real arrears – overdue customer orders and a real priority.

• Virtual arrears – caused by longstanding customer orders without any urgency
on them. These virtual arrears are orders which the customer does not really
want, or where they have bought an alternative instead and not cancelled the
order for the original item.
Outstanding orders need regular review to ensure that virtual orders are either
rescheduled or removed. If virtual arrears build up, they have a distorting effect on
the back order measurement.
2.C Demand management
2.C.1 Order processing

Stock should, of course, fulfil customer demand. In practice this includes both
having stock available and despatching it.
Arrears can be measured in several other ways. The physical handling of the
stock is a stores operation responsibility, while in many companies the actual order
processing forms part of the stock administration function. Demand management
consists of a series of steps:
a) Order receipt;
b) Order processing;
c) Estimating delivery times to customers;
Customer service 25
d) Production of despatch information;
e) Providing the items (picking, packing and despatch);
f) Customer feedback and customer care.
a) Order receipt
This is the initial compilation of orders into a form that can be used to despatch
the goods at the right time to the customer. This information must be
• complete;
• specific;
• accurate.
The information can arrive by post, telephone, electronically or by word of

mouth. A first requirement is that the company has a good item identification sys-
tem that enables sufficiently unique identification to satisfy a discerning customer.
This has to be formalised into data for processing.
Unique product coding is essential.
Stock control requires an identification code for each item, ideally provided by
the customer. To assist this, companies have website order processing and verifi-
cation or they provide customers with catalogues and order forms to enable cus-
tomers to specify precisely. It used to be customary for counter sales customers of
all types not to know code numbers, and to rely on stores personnel to remember
them. This situation is not acceptable, especially for personnel within the same
business – if staff want the correct item, then they should expect to specify their
requirement properly. An order form not only enables the correct data to be
collected, it also presents it in a standard format, which can be arranged to match
computer input or stores’ picking procedures (this improves reliability of data –
see Pokayoke technique, section 5.C).
b) Order processing
This is the entry of the order into the system, which can be integrated with the
order receipt process (as described in the previous section). Order processing
transfers the customer order into company information reliably, speedily and with-
out errors. This process needs to be fast, simple and, above all, accurate. If the
customers are entering the data, it should be simple for them to identify what they
want and organise payment. For other demand methods (text, mail, verbal), the
information should be transferred directly into the system immediately, to avoid
errors and delays. Part of the order acceptance process is checking the customer
against any outstanding debts or reasons for rejecting the order. Where printed
26 Customer service
customer confirmations are required, these should output from the system once
the delivery time has been ascertained. For non-computer applications, the source
document (or a copy of it) should be used for stock picking.
Mistake-proof the data input process.
Demand can be considered in two types:
1 Routine order processing – demand that is easy to fulfil with inventory avail-
able at the time required.
2 Demand requiring management decision – any demand which exceeds
current stock or planned output (e.g. the master schedule in Manufactur-
ing), depletes inventory excessively (e.g. a large single order for a popular
item).
Customer orders received via the internet (website or Electronic Data Inter-
change (EDI)) require no work to input into the system, except when it falls into
the second type – requiring a management decision. The system should provide
exception reporting to highlight this. There then should be a decision made to
either
• delay the delivery, or part ship; or

• reject the order (particularly for non-standard demands); or
• reallocate existing product (let someone else down).
Or, as a last resort, modify the plan (expedite supply) with the consequent
knock-on effects causing much re-planning and cost.
The inventory management has to make a decision on how to deal with each
situation: failure to decide, and to communicate the decision, results in poor
perceived availability, lack of customer trust and loss of competitiveness.
Criteria can be set up in the operating system so that the majority can be
accepted without question, and an acceptable workload is maintained. The stock
control parameters for establishing these criteria are discussed later.
c) Estimating delivery times to customers

This is the activity which causes a high level of customer confidence to be gained
or lost. Promises have to reflect the real stock and replenishment situation or else
customer service will be poor (see section 2.B.3). From an inventory perspective,
it is better to be pessimistic about delivery times; however, if there is a sales team
involved in the process optimistic lead times may result and credibility may be lost
(techniques are discussed in section 2.D).
Customer service 27
Reliability is usually more valuable than quickness.
d) Production of despatch information

Despatch information will consist of two stages:
1 picking and delivery note production, and

2 customer invoicing.
These stages should be carried out sequentially. The financial (invoicing) system
should not be allowed to interfere with the stock movement (despatch) system, as
invoicing is not part of the stock control system. Two alternative paperwork
processes can be used. Where stock is picked immediately on demand, a delivery
note can be prepared as soon as the order is processed. In this case they are not
influenced by invoicing value targets.
A. Order processing with B. Order processing with

accurate stock records inaccurate stock records
Customer order receipt Customer order receipt
Check credit worthiness Check credit worthiness
Issue picking list

Issue pick/ Inform
despatch notes customer
Pick and confirm availability
Pick and despatch note

Create Identify
despatch note Shortages
Send despatch Occasional
note with goods Discrepancy
Inform Part ship with

customer despatch note
Alter stock record
and record arrears
Alter stock record

and record arrears
Expedite Resubmit order
supply awaiting stock
Expedite Resubmit order
supply awaiting stock
Figure 2.8 Order processing options.

28 Customer service
If stock records are accurate, delivery information can be prepared from the
customer demand input as shown in Figure 2.8A. However, if stock records are
somewhat inaccurate, a safer procedure is the one shown in Figure 2.8B. This
usually causes extra work. An underlying principle is that the stock record shows
exactly what is currently in the stock location at all times, so any paperwork pro-
cessing delays really cause problems.
e) Providing the items

Monitoring delivery dates can become complex. Providing too many management
controls is confusing and dilutes the important ones. The difference between the
customers’ request for delivery (option 2) and expected delivery dates (option 1),
is their perception of lead time.
The physical processes have to be carried out efficiently within the warehouse,
and the aim is to provide the correct items, at the correct time, in the correct
quantity, in the best presentation for the user. In fact, it is better to exceed the
customers’ expectations. This means understanding the customers’ needs and
attention to their individual details.
f) Customer feedback
To report on progress and to keep the customer feeling involved in the supply is
very important from a customer relations standpoint (see section 2.A). Customers
often require acknowledgement of order and confirmation of price as well as a
despatch note and invoice. This confirmation should form part of the normal
information processing procedures.
2.D Consuming forecast demand and supply lead time

In almost all situations, business works to a forecast until the real customer orders
arrive. However, real demand rarely matches the forecast and the question is raised
as to whether an actual order received is really included in the forecast.
Consuming the forecast = using up some forecast demand by

replacing it with customer orders.
Ideally, a company forecasts demand accurately and makes that quantity avail-
able. As customer orders arrive, available quantity (or operating capacity) is allo-
cated. When the quantity for the current period has all been allocated, then
customers have to wait until the next period. This situation is shown in Figure 2.9.
Here, the forecast in periods 1 and 2 (shown as the curve) is fully allocated and
there is demand for periods 3 and 4 which have already been received. The demand
in periods 5 and 6 is all forecasts as no orders have yet been received. Orders are
constantly being received and product is allocated to them. This consumes the
forecast quantity and builds up the allocations.
Customer service 29
Forecast = sales orders expected

Maximum
limit
Current
Sales orders
Order New orders

Book received
1 2 3 4 5 6
Time (in days or weeks)
Figure 2.9 Consuming the forecast.
A company that accepts orders by ‘consuming the forecast’ does not allocate
more than has been provisioned. Since forecasts are somewhat inaccurate, addi-
tional flexibility can be achieved through:
• safety stock to meet high demand, or

• extra supply sources available to meet peaks, or
• resources reserved for emergency supply (e.g. allocation of only 90% of
capacity in manufacture, leaving 10% for last minute orders).
The penalty for agreeing to fulfil more orders than the forecast and buffer is
poor delivery achievement, much management time spent chasing problems,
panic purchases, and extra deliveries will be required, causing costs to be high.
In the long term, this gives a reputation of poor delivery and loss of market
share.
Limit customer orders to the achievable capacity.
Improvement in forecasting is the key to efficiency so that customer demand

can be satisfied reliably with changing plans and wasting time and effort in doing
this is avoided.
The differences between forecast and actual demand are measures of the quality
of the forecast (see section 12.E). Actions arising from customer order processing,
besides accepting the customers’ requests, could be to:
• arrange increase or decrease in stock levels;

• bring forward or delay supply;
• offer earlier or later delivery date to customers;
• reject order, or offer alternative item;
• provide part shipment.
30 Customer service
Example
A busy packaging company had a Master Schedule (see Chapter 14C)
for the next 6 weeks showing the planned deliveries to customers. One of
the sales team rushed back with an order from a new customer and has
promised delivery in 3 weeks to get the sale. The planner observed that
there was neither enough people nor materials to fulfil this order. However,
appeals to the Managing Director (an ex sales person) resulted in the
order being accepted from this ‘new customer with great potential’.
The result was that other customers were contacted to attempt to
delay their orders, some temporary staff were employed, suppliers were
contacted to expedite deliveries, and plans were redrawn through a
series of acrimonious management meetings.
The outcome was that many customers were upset by the delays and
part shipments and some found alternative suppliers. The new customer
placed orders sporadically (with the ‘normal’ 3 week lead time) whereas
everyone else was allowing 6 or more weeks, and the management
spend half the time progressing customer priorities.
2.D.1 Estimating delivery times

For supply-to-order items, and for stock-outs, an estimate of the delivery date is
usually required. Sales personnel are often not aware of the supply position and
the order book, and so they can only give a blanket estimate of delivery. Stock
Control, or Purchasing, do have the information, and can provide dates that are
based on the actual current situation.
It is the responsibility of inventory management to provide delivery and avail-
ability information in support of customers’ orders. The time taken to provide an
item to a customer depends on the balance between stock currently held and on
order and the quantity already allocated to customer requirements (see Chapter 12).
If the order being accepted is allowed to use items already allocated to other
customers, then the order can be fulfilled sooner. In general, the priority
sequence is ‘first come first served’, but there are often cases where certain cus-
tomers gain priority. Sales may consider that promising the items independent
of ability to deliver will secure orders and keep inventory control people on their
toes. This can lead to a mistrust of the optimistic delivery promises made, and
poorer customer service. The solution to this situation is for inventory manage-
ment to provide customer delivery dates using priority rules specified in advance
by Marketing. Where there are exceptional circumstances, then priorities can be
altered within the total availability profile. The request to ‘ring the supplier and
get a faster delivery’, should be dismissed on all but the odd occasion, since it
usually results from sales people not being able to negotiate successfully.
Customer service 31
Dates which can be considered as appropriate are:
I When the customer placed the order.

II When the customer requested delivery.
III When the customer needed the item (usage date).
IV When delivery was originally promised.
V When delivery is now promised.
VI When the item is despatched.
VII When the item is received by the customer.
And it is also useful to know:
I Who changed the delivery date.

II How many times it has been changed.
III When the delivery date was changed.
The importance of these dates varies from one market to another and one company
to another. There are so many dates and potential monitors that companies have to
choose which ones are important in their own circumstances. Providing too many
management controls is confusing and dilutes the important ones. The prime
controls would be delivery on time: dates ‘VII–II’ and ‘VII–IV’ (see section 2.B).
Decide carefully how to measure ‘on time’ delivery.
The difference between the customers’ request for delivery and expected deliv-
ery dates (II–I) is their perception of lead time. This may be a general indication
of market expectations. The difference between the customer order date and the
date they received the goods (VII–I) is the actual customer lead time. Unfortu-
nately, most companies measure their despatch dates (VI) rather than the customer
receipt dates (VII), so there could be a difference between the two views of deliv-
ery performance. Many carriers now have sophisticated tracking systems that
have the potential to provide suppliers with customer receipt information to use in
their control systems.
Measure service from the customers’ point of view.
Customers may leave an interval between their requested supply date (II) and
the time they actually need the items (III) to compensate for delivery problems. If
information is available to monitor this, it can be used to improve collaboration
between supplier and customer. The monitoring of delivery performance, as with
32 Customer service
all other inventory monitoring systems, exists to provide the best factual informa-
tion for management and control. Monitoring systems should therefore differenti-
ate between when the customer wanted the goods (II), when the delivery was
originally promised (IV) and when it is now promised (V). It is useful to measure
credibility (VII–IV) to ensure that the supply systems are working correctly.
Measuring performance against current delivery promise (VI–V) can fool a com-
pany into thinking falsely that it is performing well. The current delivery promise
may bear no relation to when the customer requires the item, and may be the latest
in a succession of promises as the delivery date slips back. The estimated delivery
date can often be adjusted so that 100% performance appears to be achieved.
Therefore, monitoring success against current delivery promise is not likely to be
the most reliable measure of effectiveness. Monitoring of slippage – the number
of times the delivery promise has changed (VIII) – can also be revealing in show-
ing whether the company is providing customers with good information and ser-
vice. However, the reason for late delivery against the original plan could be simply
that the customer does not require the goods until later. It is therefore useful to
examine (IX) what proportion of the changes are due to poor supply and how
many are due to customers changing their mind.
Key points
• Find out how to delight the customer – go for added value.
• Measuring availability is essential for success (more important than
financial results).
• Choose an availability measurement which suits the typical customer.
• Reliability is more important than speed for repetitive demand.
• Ensure that demand processing procedures are effective.
• Keep the customer informed and involved.
Note
1 If the timing and amount of demand is certain (‘dependent demand’) and the supply
completely reliable, then 100% is theoretically possible.
3 Shaping inventory
• Use of Pareto Analysis to save time and give control.

• ABC for managing inventory.
• Stock cover – a monitoring tool.
• Minimising effort and inventory value.
3.A Using Pareto Analysis for control
3.A.1. Applying effective control

Pareto Analysis is a very effective way to achieve maximum effectiveness with
minimum effort. It focuses management time on the most important issues in any
part of the business, and it is the basic tool for managing inventory to achieve
conflicting objectives of ‘best availability with least inventory and least time con-
trolling it’. In many situations, there are far too many different products to manage
them individually and Pareto Analysis provides the solution.
Inventory management is not about the unit cost of items. The aim is to control
the stock value (unit cost × quantity in stock) and Pareto Analysis is a direct
approach to this. It is not primarily aimed at different levels of customer service,
but mostly about ‘how to achieve the required availability with a minimum of
stock value and personal effort’. The basic principle for inventory management is
therefore to have tight control on the items where the turnover value is high and
not to bother too much about the inventory value of low turnover items (still
ensuring the same availability). For example, an item where the annual turnover is
£104,000 could cause a high stock value if the stock is six weeks’ worth (£12,000),
whereas if the turnover is £104, then six weeks’ worth of stock is only £12, which
is not a problem (except that there are usually many low turnover items, so the
cumulative effect is also high inventory value). Inventory managers apply tight
personal control to the important few, whilst having effective systems to ensure
good results for the rest.
A store contains items ranging from main products to cheap items (e.g. labels,
washers, etc.) with a stock record for each. High turnover value items need to be
34 Shaping inventory
controlled closely, whereas minor items need not be treated as carefully. To con-
trol the resources of the company most effectively, effort and controls should be
biased towards high cost areas. Pareto Analysis formalises this approach. It states
that the majority of the effect is produced by a small proportion of the causes
(typically 80% of the effect is due to 20% of the causes). Therefore, Pareto’s Law,
illustrated in Figure 3.1, is also called the 80–20 rule. It can be used to direct effort
in most situations, including:
• Suppliers (expenditure per supplier)

• Purchased item costs (purchased quantity × unit cost)
• Stock value (stock quantity × unit cost)
• Profits (annual profit per line)
• Customer accounts (debt per customer)
• Management time spent (minutes per job)
• Warehouse space (size × stock quantity)
• Warehouse movements (transactions per line)
• Equipment breakdown (downtime × frequency per cause)
(The evaluation methods are in the brackets.) Focusing on the major contributors
can have a major impact in minimum time.
The shape of the Pareto curve (Figure 3.1) arises from the range of volumes
and values combined in a statistical distribution; 80% of the effects come
from the top 20% of the causes, but this should not be the focus as practi-
tioners are more likely to use other points on the graph (65% from 10% or
100%
90%
80%
70%
60%
Effect
50%
40%
30%
20%
10%
0%
10 20 30 40 50 60 70 80 90 100
Cause
Figure 3.1 Pareto curve.

Shaping inventory 35
95% from 50%). The curve does not always give an 80/20 relationship, but
this does not affect the principles of applying Pareto Analysis, especially to
inventory management.
Two important ways of using Pareto Analysis are based on:
1 Turnover (demand rate × unit cost). Providing long-term inventory balance

and called ‘ABC analysis’.
2 Current stock (stock quantity × unit cost). Altering the current stock levels.
If the immediate aim is to reduce stock investment, then studying the stock of
low value items is unlikely to be the best place to start unless the sales volume is
very large. If service is the aim, then attention to a few fast-moving lines often
provides the bulk of the improvement required.
Use Pareto Analysis to save time and get results.
The application of this to stores stock control means that 80% of the total stock
value is made up from 20% of the total stock items.
80% of stock is in 20% of lines – so reduce the high stock values.
The other 80% of stock items contribute only 20% to the total inventory value
(shown in Figure 3.1). In a stock reduction exercise, the majority of our cost saving
will be gained by decreasing stocks of those few major lines.
Example
There was a stock of 12,000 types of items in the electrical stores area.
Pareto’s Law shows that for a stock value of £800,000, only 2,400 items
accounted for £640,000 of inventory. The remaining 9,600 items were
worth only £160,000. Therefore, by concentrating on the 2,400, control
over the total value was tightened. 2,400 were rather too many to review
individually on a weekly basis. As the Pareto curve (Figure 3.1) shows
that 5% of items account for 55% of cost, then 600 items contributed
£442,000 to the total stock costs. By working on these 600 items care-
fully, the overall stock value was controlled and decreased.
3.B ABC analysis
The current stock does not necessarily show which items are important for the
business. In fact, there may be some important items where the current level of
stock is low because Stores are awaiting an impending delivery. On the other hand,
some items may have a high stock value simply because no-one is buying these.
It is therefore usual to rank the items according to the annual turnover (measured
at cost, not sales value). The annual turnover is given by
Turnover = Annual Usage × Unit Cost
This simple formula may have to be modified in the case of new or obsolescent
items to reflect the future expected demand rate rather than the historical one.
The original definition for Pareto classification therefore was ‘forecast demand
rate’, not ‘annual usage’. It is not too important whether the unit cost is the stan-
dard cost, latest cost or an average as long as it is consistent across all items.
Pareto Analysis shows that 80% of the turnover is for 20% of the moving items.
There is often a number of items in stock for which the demand is zero, and there-
fore not included in this turnover analysis. For some businesses, the 80/20 rule is
not obeyed exactly, but Pareto Analysis should still be used. The traditional simple
approach is first to classify the items by turnover value into three classes: A, B and C.
In practice, the classification should be:
A class = 10% of stock lines, giving 65% of turnover

B class = 20% of stock lines, giving 25% of turnover
C class = 70% of stock lines, giving 10% of turnover
This is illustrated in Figure 3.2.

In much of the literature, 20% of stock is allocated to A class. For businesses
with more than 500 different stock lines, it is sensible to reduce the number of
A class items to focus control and reduce the workload, so the classification in
Figure 3.2 is used.
ABC analysis is an excellent technique for achieving objectives.
It is important to ensure that ABC analysis is based on turnover, but less import-
ant that the exact percentages are adhered to. Dividing turnover into more classes
(ABCDEF) can tighten control over supply quantities, and this was illustrated by
Relph and Milner.1 Many companies add a D class for a large number of very low
turnover value items. This enables the number of stock lines included in the A, B
and C classes to be reduced to a manageable size. A further development of the
Pareto approach for the supply chain is the ‘9-box model’, discussed in Chapter 15.
100
90
80
70
60
Effect
50
40
30
20
A B C
10
0
10 20 30 40 50 60 70 80 90 100
Cause
Figure 3.2 ABC analysis.
The purpose of ABC analysis is not to provide different types of service, but
to provide service with the least amount of cost and effort. Different systems of
control are used for the three categories of stock. As the A items carry the most
value, accurate methods are required to control them. On the other hand, the C
items are low turnover value, but form the bulk of the inventory. For these, the
main requirement is to ensure that stock is available to meet demand.
A class – there are not many, so control tightly.
The control requirements for each category are shown in Figure 3.3. Category A
items have a disproportionate amount of time and effort spent on them. They should
be controlled tightly using expertise, market intelligence, product knowledge and
system generated forecasts to maintain stock at the lowest appropriate level.
B class – let the system manage these.
For category B items, computerised techniques are most appropriate. The num-
ber of items involved and the lower values make it a waste of time to use specialist
skill that could be working on category A. Software can maintain control through
statistics using complex calculations from historical forecasting algorithms.
Characteristics Policy Method

A items Tight control Frequent monitoring
Personal supervision Accurate records
Few items
Communication Sophisticated forecasting
Most of turnover
JIT approach – Service level policy
balanced safety stock
B items Lean stock policy Rely on sophisticated
Use classic stock control method
Important items
Fast appraisal method Calculated safety stocks
Significant turnover
Manage by exception Limit order value
Computerised
management and
exception reporting
C items Minimum supervision Simple system
Supply-to-order where possible Avoid stock-outs and
Many items
Large orders excess
Low turnover value
Zero or high safety stock policy Infrequent ordering
(Few movements or
Automatic system
low value items)
Figure 3.3 ABC inventory control.
The use of management by exception is also important for B class where history
is not a good guide. The manager can then intervene when the system cannot cope
with the situation (Chapter 12).
The minor sales items, category C, should be controlled by a simple system
which enables supply to be obtained with a minimum of administration. For C
items, the key is a reliability of supply, so holding marginally more inventory
means less administration and frees up time to work on A items. However, because
most of stock items are C class, the cumulative C class inventory value can be
high. Separating C class by unit cost and usage (Chapter 15, 9-box model), is
advantageous. Control through a simple two-bin system is appropriate if reliable.
C class – do not take risks or be lazy. High but not excessive inventory.
The most effective stock control systems are based upon ABC analysis com-
bined in a common-sense manner with the other techniques discussed below. ABC
analysis is also used as the basis for perpetual inventory stores control where
annual stocktaking is avoided by routine counting of a few stock parts each week.
Classifying the turnover value using Pareto Analysis enables a business to orga-
nise the supply of goods. This is the key application, but in some situations the
technique can also be used to manage availability for customers where better
availability can be offered for major turnover or profit items.
An example of Pareto Analysis in action is given in Figures 3.4–3.7. Figure 3.4
shows a number of different stock items, their unit cost and annual usage in terms
of quantity of value.
Item Annual usage Unit cost Annual Annual Rank
(units) (£) turnover (£) turnover (%)
A12 21 7 147 2.1 5
B23 105 11 1,155 16.3 2
C34 2 15 30 0.4 10
D45 50 5 250 3.5 4
E56 9 14 126 1.8 6
F67 394 12 4,728 66.6 1
G78 5 8 40 0.6 9
H89 500 1 500 7.0 3
I90 11 4 44 0.6 8
J01 3 25 75 1.1 7
Total 7,095 100
Figure 3.4 Example of Pareto Analysis. Re-arrange
Item Annual Unit Annual Annual Rank Class Cumulative

usage (units) cost (£) turnover (£) turnover (%) percentage
F67 394 12 4,728 66.6 1 A 66.6
B23 105 11 1,155 16.3 2 B 82.9
H89 500 1 500 7.0 3 B 89.9
D45 50 5 250 3.5 4 C 93.4
A12 21 7 147 2.1 5 C 95.5
E56 9 14 126 1.8 6 C 97.3
J01 3 25 75 1.1 7 C 98.4
I90 11 4 44 0.6 8 C 99.0
G78 5 8 40 0.6 9 C 99.6
C34 2 15 30 0.4 10 C 100
Total 7,095 100 Classify
Figure 3.5 Classification by usage value.
Classification Percentage of items* Percentage of value Value per class (£)

A 10.0 66.6 4,728
B 20.0 23.3 1,655
C 70.0 10.1 712
Total 100.0 100.0 7,095
Figure 3.6 Summary of ABC analysis.

*
ABC analysis is carried out only for items with usage.
100
Percentage turnover 80
Cumulative
60
40
20
0
0 2 4 6 8 10
Sequence of items
Figure 3.7 The cumulative Pareto curve.
The process is then:
• Sort the items in order of size of decreasing annual turnover value (shown in
Figure 3.5).
• Calculate the percentages of turnover and the cumulative turnover values.
• Classify the top 10% as A, the next 20% as B and the rest as C, except those
with zero turnover which are X (supply only to cover an outstanding customer
order).
Figure 3.6 summarises the results. It shows that the seven items in C class have
a combined turnover of £712, whereas the only item in A class has a turnover of
£4,728. It is obvious where the inventory controller should put effort into keeping
low stocks whilst providing the same excellent customer service. At the same
time, the availability of C class items is just as important as A class.
3.B.1. Saving time

Applying the Pareto principle is a way of balancing inventory, stock availability and
critical resource spent on each item. It can be used for focusing control to achieve
best results in most situations. The critical resource for all inventory controllers is
time. They have a large amount of information required for tight control and a wide
variety and quantity of stock. Using Pareto, 80% of time is spent doing 20% of the
jobs and a significant time saving can be made if a small reduction can be achieved
doing these few. They may be very frequent short jobs (such as loading stores
issues into a system) or infrequent, long-winded jobs (such as writing a major
management report). Analysing and organising time leads to effective control.
3.B.2 Practical considerations when using ABC analysis
The ABC classification is a simple tool to enable the stock manager to control a
lot of items in a limited amount of time. It is one of the most powerful tools
employed to reduce stock value and to decrease the workload of busy inventory
managers and purchasers. Experience with this technique over a large number of
companies has suggested that there are some pitfalls and some practical ways to
circumvent them. The situations which arise are typically:
1 too many A class items;

2 large numbers of lines (add D class);
3 non-moving lines (add O class);
4 fixed stock level items (add F class);
5 non-stock items.
Use ABC to save time.
3.B.3 Number of A class items

A class items are supposed to be reviewed on a daily basis or a weekly basis at
least. To be practical, there has to be an upper limit of, say, 300 A class items per
inventory planner. Even where the planner manages many thousands of items, the
number of A class items should be kept small.
3.B.4 D class items

For normal inventories of 3,000 lines, the ABC classification works well. Where
there are more than 10,000 stock lines, there the classification system should be
modified so that the vast bulk of low turnover value items are dumped into a further
class, D. The D class contain the lowest turnover lines, say 50% of the active item
numbers which contribute only 2–3% of the total turnover. The ABCD classifica-
tion now is as follows:
• A class 5% of moving lines (300 per planner maximum)

• B class 10% of moving lines
• C class 35% of moving lines
• D class 50% of moving lines.
The classification can alternatively be carried out by turnover value; for example:
• A class 45% of turnover

• B class 30% of turnover
• C class 22% of turnover
• D class 3% of turnover.
These classifications can be varied to suit the exact shape of the Pareto distribu-
tion for the inventory to be managed. The principles of use of the technique do not
change even if the distribution is not 80/20.
3.B.5 Fixed classification, F

So far in the discussion the moving inventory has been classified, and the non-
movers have been ignored.
As Pareto Analysis is to be used for replenishment, there are items where the
stock level should not respond to usage rate. It is important for all other items that
inventory parameters are altered regularly so that stock levels reflect current
requirements. Companies do sometimes have a need to keep non-moving items, or
items where the movement is so slow that they appear as non-movers in the record-
ing systems.
For example, the employment of two new maintenance fitters in a factory
requires them to be kitted out with protective clothing, tools, tool boxes and a
variety of items. If this is a one-off happening, not a regular occurrence, the provi-
sioning of these items should not increase as a result, so the stock level of these
items should be fixed. By putting these items in a separate classification (say, F),
they can be segregated and identified. The system can then avoid updating the
stock level parameters from classification F.
Use ABCOFZ as a practical control method.
3.B.6 Non-stock items, Z

The decision as to which items are in the stock range is an arbitrary one and depends
on the particular inventory and marketing policy. Customers do not usually con-
sider their requirements as ‘stock’ or ‘non-stock’ items and the difference to them
is only in the lead time provided by the supplier. There is a continual shift with
non-stock items added to the stock range and existing stock items being deleted.
For this reason, it is useful to include non-stock items in the Pareto turnover analysis.
By adding another class (Z), the non-stock items can be included. This enables
items to be included where there is remaining stock from a line being deleted from
the stock range. The specification for this class is that no stock is ordered from
suppliers unless the stock cover is negative (i.e. there is a customer requirement but
no stock). This avoids the embarrassing auto-ordering of obsolete items.
3.B.7 Obsolete items, X

Obsolete items will probably be subject to the stock cleansing discussed in Chapter 4.
In the interim they should be provided with a separate classification. Normally, they
are given class X, which signifies that they should not be ordered again. Items that are
superseded can then be put into this class, along with those where demand has ceased
naturally, or the business decides not to provide them.
It is very useful to classify all goods in the inventory management system as it
provides unified records identification and makes management and analysis easy.
3.C Stock cover
3.C.1 Turnover of stock

The current stock levels in the various stores throughout the company may not all
be at ideal levels, as we have seen. The purpose of controlling the inventory is to
drive stocks towards their proper level, which is determined by the characteristics
of supply and demand patterns. The major factors are:
• average demand rate;

• variability of demand;
• supply frequency;
• customer delivery time allowed;
• supply lead time.
There are also practical considerations, such as:
• reliability of the supplier;

• criticality of the item;
• availability of item from other sources;
• batch size of demand.
The concept of ‘balance’ is most important in ensuring that the maximum service
is produced from a minimum of stockholding cost.
The best level of service will be provided if there is an equal chance of all items
being available for the customer. High stocks of one item and no stocks of another
will reduce overall availability and increase inventory cost. The percentage avail-
ability target (Chapter 2) should therefore be the same for all items, although it
may vary between markets.
If the cost of managing the inventory is also considered, then the value of items
being ordered and controlled becomes important. Valuable inventory management
resources should be confined to the most cost effective jobs. This means that time
should be spent on high turnover value items and not wasted on items whose value
is insignificant. The best balance of inventory leads to an optimisation of costs and
service over the full range of stock lines within the time available.
The inventory performance of each item can be monitored using a figure of
merit for stock balance. This is the ‘stock cover’, which is defined as:
Current Stock × 52
Stock Cover =
Forecast Annual Usage
Item code Stock (units) Annual usage (units) Stock cover (weeks)
1P1 250 2,000 6.50
1P2 700 1,625 22.40
1P3 500 400 65.00
1P4 15 1,000 0.78
1P5 20 25 41.60
1P6 40 250 8.32
1P7 500 200 130.00
1P8 8 400 1.04
1P9 6 40 7.80
1P10 65 20 169.00
Figure 3.8 Stock cover.
This gives the result in terms of weeks on hand. Similar, but not quite as rel-
evant, results can be obtained using historical averages, which are often more
convenient, or shorter-term forecasts, changing the ‘52’ to suit (Chapters 11–13
on forecasting).
It is important to realise that stock cover (weeks of stock) is a very bad way to
set safety stock levels, and many inexperienced managers fall into this trap.
A sample of stock items is shown in Figure 3.8. Which of these items requires
attention first?
Stock cover gives an insight into the priority for action. It is not an infallible
guide, but it does indicate where review is required. This shows clearly that 1P4
and 1P8 are at risk, where the stock cover is small. However, the controller may
know that there are no immediate requirements or that delivery is imminent. At the
other end of the scale 1P3, 1P7 and 1P10 have over a year’s worth of stock, so
means of reducing this level will have to be found.
Use stock cover for analysis but not control.
Stock cover shows whether the stock is ‘in the right ballpark’: it directs atten-
tion to those items that are obviously well outside acceptable stock levels. It keeps
the shape of the inventory reasonable. Stock cover is used by inventory controllers
because it is easily understood in terms of usage rates and lead times. The value
shows the time ahead when the stock will run out (at average usage rate).
As well as being a crude analysis tool for each stock item, stock cover is also an
important tool for measuring total inventory investment. Overall stock cover is
calculated from the total value of stock divided by the annual issue value and
multiplied by 52 to change the answer to weeks. In many distribution stores, the
answer is between one and eight weeks. In fast-moving consumer goods, one to
eight days is more appropriate and, with kanban systems, one to eight hours.
Financial managers are often more interested in the use of funds, and therefore
measure the effectiveness of inventory management using ‘stock turnover’ or
‘stockturn’. This is just the reciprocal of the stock cover, taken on a value basis for
the complete stockholding.
Value of Annual Usage

Stock Turnover =
Value of Stock
This calculation gives the number of times the stock would be used up per year.
Example of stock turn calculation

Value of stock in the stores is £150,000.
Issues for the last 12 months amount to £900,000.
Stockturn is therefore 900,000 ÷ 150,000 = 6.
This means that the stock value would be used up completely six
times per year so that the stock cover for the total stock will be two
months, or by the stock cover calculation
Stock Cover = 150,000 × 52/900,000 = 8.67 weeks.
Stock turn is based on historical data and is used for financial reporting. Stock
cover is an inventory management tool for planning stockholding and can be
based on known data and forecast usage rate, so that the stock will meet the
expected demand for the item. When the stock level is being assessed for account-
ing purposes, the ratio uses the historical usage rate, which enables a conservative
view to be taken of the stock level. Although this sometimes leads to a divergence
of views on the necessary stockholding, the assessed future demand should always
be used when controlling stock or placing orders.
3.C.2 Setting stock targets

For good stock balance, the stock cover of all the items in an ABC class should
have similar value. In practice, differences in the variability of usage and the order
cycles leads to a range of acceptable values for the items. Stock cover should not be
used for working out re-order levels – there are proper, accurate ways of doing this
(Chapter 8). Stock cover ratios can be used to calculate the broad ranges of weeks’
cover which would be needed for inventory items. For instance, it is unlikely that
more than 26 weeks’ worth of stock is planned for any inventory item, therefore a
figure of more than 26 could be the boundary between ‘OK’ and ‘needs attention’.
As inventory control should be tightest for the A class items, these are the ones
that can be controlled down to lower stock cover figures (leading to the paradoxi-
cal situation of holding lowest stocks of the best sellers – try explaining that to the
sales people!), whereas extra stock of the minor C class items adds little to stock
value and significantly reduces the workload of controlling.
In practice, the stocks could have control limits to avoid extremes of inventory,
and an allowable stock cover range can therefore be set by the ABC inventory
classes in a ratio which is theoretically 1:3:7. An illustration of the acceptable
ranges for category A, B and C items would typically be:
• A class between 1 and 4 weeks

• B class between 2 and 8 weeks
• C class between 3 and 20 weeks.
Stock cover (weeks of stock) for all the items in the category should lie in the
ranges shown (Figure 3.9). For the vast majority of companies, the horizontal
scale in Figure 3.9 is ‘weeks’. The exceptions to this rule exist for perishable food,
which uses ‘days’, and aerospace spares that can only manage ‘months’. Of course,
where kanban supply can be arranged (predominantly for A class), stock cover is
less than a day. The inventory manager should identify comfortable limits intui-
tively, try to control within these, and then reduce them.
Figure 3.9 shows the theoretical stock cover for well controlled items in each class
(note that the horizontal scale is logarithmic). The ideal and typical results are shown
in Figure 3.10. The theoretical curve allows for variability of demand, some usage of
safety stock for items and slow movers. This causes some items to fall naturally out-
side the expected class limits. The shape of the actual population curve results from
pressure on stock value. Problem items have high stock cover ratios, and since the
stock value has to be limited, the stock of faster usage items is squeezed to compen-
sate and reduce inventory value. This can have a detrimental effect on availability.
The actual profiles in Figure 3.10 are typical, even where inventory manage-
ment practices are relatively good. It is good to create Figure 3.10 in practice.
Number of
items
C
A
1 2 3 4 8 20
Figure 3.9 Ranges of stock cover.

Actual
Number of Lines
Theory
0 5 10 15 20 25 30 35
Weeks (or Months)
Figure 3.10 Expected and actual stock profiles.
3.11a
Class Number Value of Stock Average Average delivery Number of
of items turnover cover stock value frequency deliveries
% % Weeks Weeks Weeks Annually
A 10 65 1.5 0.975 3 173
B 20 25 3 0.75 4 260
C 70 10 6 0.6 12 303
Total delivery quantity stock 2.325 737
19% Per 100 total lines
3.11b
A 10 65 0.5 0.325 1 520
B 20 25 3 0.75 4 260
C 70 10 8 0.8 16 228
3.11c
A 10 65 1 0.65 2 260
B 20 25 3 0.75 4 260
C 70 10 8 0.8 16 228
Figure 3.11 Minimising stock value the easy way.

The weeks’ stock cover shown in Figure 3.9 indicates how long it would
typically take to use up the inventory. If the stock reduction is to be made through
C class items, the time taken to reduce the levels will be in excess of 20 weeks.
For A class items, the whole process should be completed within four weeks.
3.D Pareto stock balance

A canny inventory manager applies Pareto Analysis and stock turn to fool the
accountants that they have wonderful control without doing too much work.
The use of these two techniques (ABC analysis and stock cover) together is
fundamental in forming the basis of the stockholding policy. An example will
illustrate the principle.
Example
A business needs to reduce inventory investment. It has just 100 different
lines and has the typical stock cover for the A, B and C classes shown in
Figure 3.11.
The A, B and C classes have average stock cover of 1.5, 3 and
6 weeks, and therefore constitute a well-controlled inventory with a total
stock cover of 2.325 weeks. If this stock cover appears to be over gen-
erous or small, then adjust the units of measure – day, fortnight or month –
but keep the ratios. The classes contribute 65%, 25% and 10% to
inventory value, respectively. This means that the total stock values
arising from the A, B and C classes are similar, with the 10% of A class
items producing more inventory value than the 70% of items compris-
ing C class (Figure 3.11a: 0.975 of the total weeks’ value against 0.6).
One way of decreasing the stock value is to get smaller batches more
frequently (Figure 3.11b), and cut the highest A class inventory down to
0.5 weeks by simply increasing the supplier delivery frequency on A. This
is often easy because these are the significant purchase value lines and the
business has influence on suppliers. If at the same time the C class stock
cover is increased an average of 8 weeks (delivery frequency of about
10 weeks to allow for safety stock), the result is a reduction in stock by 19%.
This increases the number of deliveries per year from 737 to 1,008,
which may not be welcomed by the people involved. To avoid this, the
delivery frequency on A class could be reduced and the average stock
cover increased to 1 week (Figure 3.11c). Now the workload is almost
the same as it was (748 against 737) and there is still a reduction in
inventory, but only 5%. It depends on the situation of the business
whether the inventory value or workload is more important.
High stocks and poor service go hand in hand.
Tighter control of A items is the better alternative because there will be:
• fewer lines involved, so less renegotiation required;

• better chance of success because these are the highest turnover items;
• less volume of supply deliveries as a result;
• improvement that can be achieved more rapidly.
This last point can be proved by considering stock cover. Stock of A class items
lasts typically 3 weeks in Figure 3.11a, so supply changes can be made in that time
frame. For C class, the stock is typically 6 weeks’ worth so alterations in supply
patterns will come into effect much more slowly. Stock reduction through normal
consumption takes at least these respective times. As stocks are not often perfectly
balanced, the true time to reduce the stock is typically over twice this long.
Using ABC and stock cover together saves time and inventory.
3.D.1 Pareto-based supply

Supplier delivery patterns based on Pareto Analysis save stock value and manage-
ment time. This is well illustrated by the ordering cycles shown in Figure 3.11.
This inventory is 100 lines; if each line were purchased each month from the
suppliers, there would be 100 lines delivered each month (1,200 per year). The
stock cover is just over 2 weeks plus safety stock (stock goes up by 4.2 weeks’
worth on receipt and then reduces to the safety stock, so the average could be
typically 3 weeks’ worth; B items in Figure 3.11a).
Now, if the A class items were ordered twice as frequently and the C class half
as frequently, what would be the effect? Figure 3.11 illustrates the results.
• In Figure 3.11b the inventory value is reduced by a massive 19%, and it was
not bad to start off with. This has been achieved through organising the supply
of A class and relaxing the deliveries on C class.
• It has resulted in an increase in the number of deliveries, but it is still less than
ordering everything monthly. As the delivery frequency increases for the major
lines (particularly the αA in the 9-box model), this should be arranged effi-
ciently through rolling schedules, kanban or VMI (vendor managed inventory)
so the increase in deliveries is not reflected in increased workload.
Figure 3.11c maintains the number of transactions, but only gives a 5%

inventory reduction. However, reduction in effort and decrease in stock can often
be brought about at the same time. These results are typical and are borne out in
practice. If customer service is affected by this policy, it is in a positive way
because safety stock remains the same in each instance (the availability is based
on the calculations shown in Chapter 7).
Applying the approach illustrated in Figure 3.3 and setting targets including
safety stock (as shown in Figure 3.9) enables the overall inventory cost to be reduced
together with the purchasing and warehouse transactions without detracting from
customer service, and often improving it!
Non-moving stock has not been discussed in this chapter, as it is the topic for
Chapter 4.
Key points
• Pareto Analysis is a key technique for managing complex situations.
• ABC analysis provides the best tool for saving time and structuring
inventory.
• Categorise all items by a code identifying the control pattern.
• Stock cover enables inventory managers to look out for likely
problems.
• Stock cover should not be used for setting safety stock levels.
• ABC supply optimises inventory.
Note
1 Relph, G. and Milner, C. 2015, Advanced Methods for Managing Inventory within
Business Systems, UK: Kogan Page Ltd.
4 Practical methods for
reducing stockholding
• Approach to value reduction.

• Important techniques for batches and risk.
• Focusing effort and project organisation.
• Practical techniques for obsolete stock.
4.A Approach to inventory reduction

There are conflicting views about the benefits of stockholding. As business aims
to maximise service and minimise costs, inventory provides the lubrication. It
must flow. The costs of warehousing and financing should be minimised.
Those people who try slimming realise that the simple truth is:
Decrease (In Stock) = Output − Input
and there is no way round it. The options are therefore to either reduce input or
increase output or both. A normal range of stock in stores comprises fast-moving
parts, a wealth of obsolescent and special items and fluctuating quantities of other
parts – some in short supply, some over-provisioned. In most companies, there is
potential for decreasing stock by a significant amount, typically 30% over 18
months.
The first step is to estimate what causes the most stock from the classes dis-
cussed in Chapter 1. The usual problem with stockholding is the value. For each
stock line,1 the average value is:
Unit cost × Average stock
There is normally little chance of changing the unit cost significantly. However,
this should be considered, since a 10% price reduction causes a corresponding
drop in inventory value in the long term.
52 Practical methods for reducing stockholding
Safety stock
+ Delivery quantity
Usage
Delivery
quantity
Average
Stock stock
quantity
½ delivery Delivery
quantity
Average
Safety stock
stock level
Safety stock
Time (days or weeks)
Figure 4.1 Stock levels.
The average stock quantity is typically half way between maximum and minimum2
stock. The minimum should be the safety stock and the maximum occurs immedi-
ately after a delivery, so
Average stock = ½ (max. + min.)

= ½ (safety stock + delivery quantity + safety stock)
= safety stock + ½ supply delivery quantity
See Figure 4.1.

In order to reduce the average stock, therefore, there needs to be a reduction in either
a) supplier delivery quantity, or

b) safety stock.
Supplier delivery quantities are more important than lead times.
a) Supplier delivery quantity: If the quantity bought is the same as the quantity
sold, then stock is minimised. Ideally, the sale would be immediately after the
purchase (just in time), but this requires excellent forecasting. The simple
message is therefore ‘buy one at a time’ where possible. In practice, inventory
should be minimised by making the suppliers deliver small batches. People
are astounded by the policy to ‘hold the least stock of the highest turnover
value items’, but this is a consequence of good stock control. This is normally
the most fruitful way of reducing stock value.
Practical methods for reducing stockholding 53
There is often a trade-off between delivery and inventory costs. For A class,
inventory costs are greatest, whereas for C class, delivery costs are significant.
This means that batch sizes may be increased unless the supply is co-ordinated
with other items (hence the use of supplier parks for automotive supply).
Minimise the supplier delivery quantities of the A class items.
b) Safety stock: Minimising the safety stock results in worse customer service.
As stock is never balanced perfectly and the levels constantly need to be
changed, the actual availability needs to be monitored continuously. Sometimes
the safety stock far exceeds the quantity required to give excellent availability.
Achieving a higher availability level than necessary is very expensive. Therefore,
the first step in controlling the high value stocks is to ensure that the safety
stock is sufficient to provide the service required and not more. Initially this
can be done by understanding and experience. The quantitative approach is
discussed in Chapter 7. The availability for each of the high value items has to
be matched to the requirements of the market. Balancing these stocks is most
important. As the safety stock depends on the availability required, the lead
time and the variability of demand, the next topic to consider is lead time.
4.A.1 Forecasting
Safety stock is needed because forecasts are inaccurate (expect the unexpected).
Improvement in forecasting therefore reduces the need for stock. Two aspects
need considering:
1 Ensure that forecasts are updated frequently.

2 Use improved quantitative forecasting techniques.
4.A.2 Demand smoothing
Consistent demand enables stock to be low.
There is a pressure to maximise sales, and this is thought to require large orders.
Lean supply principles show that it is better to have consistent regular demand.
This is better for supply chain flow, more economic and requires less inventory.
User consumption is usually in small quantities, so incentives are required to
motivate sales personnel to agree repetitive supply instead of large batches. The
results would be smaller batch supply, easier forecasting and less safety stock,
ensuring lower stockholding.
Example: the case of a pan

The stock records for a range of materials sold by a major supplier of
aluminium were typically as follows:
Week 1 2 3 4 5 6 7 8 9 10 11 12
Demand 225 2,625 75 100 200 50 125 100 0 2,050 300 125
The items were aluminium circles (flat round sheets) in various sizes
used for making pans. The conventional stock level calculations sug-
gested very high stock levels in order to cope with the demand peaks in
weeks 2 and 10. This was introducing a high level of stock for these
items, and inflating the inventory value.
Investigation of this order pattern showed that, across the range of
products, these peaks were caused by one major customer who had an
order cycle of two or three months and this was causing the large
demands. There are several solutions to this problem:
Understand demand: Greater understanding of the demand pattern

improves the forecast. Since safety stock is required to compensate
for forecast inaccuracy, the improvement of forecasting reduces
safety stock. There is therefore the opportunity to try to smooth cus-
tomer demand by getting the customer to buy little and often.
Flatten demand: Remove sporadic demand from the stock by buying
to order and giving the customer a longer lead time.
Agree scheduled supply: It would be better to organise a continuous
supply by scheduling delivery.
4.A.3 Lead times

Shorter lead times reduce stock, but only by a minimal amount (square root factor
in the safety stock calculation). Compared with reducing the supply batch size and
improving forecasting, this is not worth doing except in two situations:
1 Supply time can be made shorter than the demand lead time (converting to
supply-to-order).
2 Sporadic demand where most inventory is safety stock (typically for slow-
moving items). This situation is similar to project management where lead
times are important.
4.A.4 Obsolete stock

Inventory reduction should not start with obsolete stock, as it is not the major
cause of stock value. Furthermore, it is often difficult to get rid of. The causes of
obsolete stock need addressing and any outstanding supply should be cancelled.
4.B The reduction project
What causes the inventory value? There is often a confusion between assessing
the inventory value and the processes for reducing it. A simple Pareto Analysis of
stock values (stock quantity × unit cost) shows where the major costs are. Start
with the item with the most stock value and continue down the stock value Pareto
curve. A target should be set so that the stock reduction for each item can be
judged against it. For example, an overall stock value reduction of 20% can be
achieved through a 25% reduction in stock of the top 20% of lines, so this is the
initial target. This is unlikely to be achieved for some items, but it is essential to
concentrate on achieving the target for the highest value lines, or the overall target
will not be met. The value underachievement on some items has to be compen-
sated by extra value savings on other lines.
A potential stock reduction target can be assessed using a metric such as IQR
(inventory quality ratio). To measure this, the total inventory is divided into Active,
Slow-Moving, Excess and Non-moving (Obsolete?), then
Active
IQR =
Total
This enables a measurement for managing assessment of the project target, but
does not identify where to put the major effort – it should normally be to work on
the major cost items, which are normally the Active type.
It is all about reducing the A class.
Obviously, the focus of a reduction project should be the type of stock with
highest value. For businesses in the distribution chain, this could be inventory
from specific suppliers or providing goods for a type of market (e.g. sale for proj-
ects, spare parts or highly variable demand). Where the whole group of items has
a high stock turn, there is likely to be good potential for decreasing inventory
value. Similarly, manufacturers have stocks of raw materials, work in process,
finished goods, machine spares and consumables. They can reduce inventory by
co-ordinating and sharing stock with their supplier and customer chains and at the
same time use lean pull systems within their operations.
4.B.1 The stock reduction process

Stock reduction is usually undertaken as a project, and concentration should be
given to achieving the objective in a short time. The first few lines in the Pareto
inventory value curve can give the major savings quickly with the least effort.
Approximately 80% of the value of stock exists in 20% of the stock lines. These
items are likely to be a mixture of fast movers with reasonably high unit value, and
high value items where there are relatively few demands per year. The way to reduce
the stock value is to concentrate on high stock value items, whether they are slow
moving or fast moving (the value of inventory is the same for each and the financial
investment is the same). In fact, it is easier to vary the inventory levels for the faster-
moving items. As far as basic Pareto classification is concerned, there is no differ-
ence between the stock for fast-moving and slow-moving items with the same stock
value. However, the approach to reducing the inventory value can differ significantly.
The stock of a fast-moving item is likely to be consumed quickly and consis-
tently through servicing a wide variety of demands from a large customer base.
The inventory profile for a fast mover is therefore large numbers of demands taking
a small proportion of the inventory. Demand is relatively stable.
High stock value fast movers can be changed quickly.
For high value items, there are relatively few issues to a smaller number of
customers. This means that the demand is more unstable for two reasons:
• As the stock lasts longer, the risk of demand changing within the stock cover
period is greater and therefore there is an enhanced safety stock and a risk of
the stock remaining as obsolete or excess.
• The demand comes from only a few customers, if any of them reduces their
offtake, this will have a much more significant effect than if there were more
customers.
It is therefore especially important to manage risks on the slow-moving A class

items by forecasting, ordering little and often, and collaborating with customers
and suppliers.
A class items are selected for special control. Apply supply scheduling where
possible and negotiate closely with the supplier. Adopt supply chain management
so that stock is only held at one level in the chain. Inventory should be pushed
back down the chain and deliveries made very frequently.
4.B.2 The stock reduction project

Specific stock reduction can be achieved through a dedicated project. Continuous
stock reduction is a background job for all inventory managers, but since insufficient
focus is given to it, arranging a project tends to be more effective. The following steps
may be taken:
1 Get an up-to-date inventory valuation and perform Pareto Analysis.

2 Take the highest value stock line and consider ways of reducing it:
a) Arrange more frequent deliveries from the supplier.
b) Improve forecasting and safety stock evaluation.
c) If the item is customer specific, review co-ordination with the customer(s).
d) Ensure supplier reliability (liaison, penalties, re-source).
e) Gain additional sales (without increasing supply or safety stock).
f) Move stock to a more advantageous part of the supply chain.
It does not matter whether this stock is considered under control, it is still high
stock value and therefore needs to be reduced.
After applying these ideas sufficiently to this item, they can be applied succes-
sively to the items along the Pareto curve.
Actual example
A company received car components from China each month (with
12 weeks’ lead time), and these are their major stock lines (average
2 weeks of delivery stock). By arranging to receive the components
weekly, their inventory reduced to ½ week delivery stock.
The aim is to make a permanent reduction in the average quantity of significant

stock value lines. The current stock of each line varies greatly since it is low
before suppliers deliver and much larger afterwards. Therefore, some A class
items may not be near the top of the current Pareto stock value curve, but they
have the potential for increasing inventory value. Scheduling small, regular deliv-
eries can make a significant reduction without disadvantaging vendors that
embrace lean principles (see Chapter 6).
Do the simple thing – start with the highest stock value line.
In a reduction project, all purchase orders should be examined to see if the pur-
chase quantity can be halved and the purchases made twice as often. The author
used this technique in practice and reduced inventory by 40% without affecting
customer service.
Avoid ordering more than (say) three months’ supply of anything. Setting a
limit ensures that the amount of slow-moving stock is reduced. Very long lead
time items will have several purchase orders outstanding simultaneously.
High stock results from low stock cover of high turnover value items and high
stock cover of some others that have been bought in bulk or where sales have
evaporated. In both cases, a major effort should be made to reduce the supply
batch size. It is often even better to find an alternative supplier at a marginally
greater unit cost.
EXCESS Stock cover

more than ‘y’ weeks
OBSOLETE
Not used for ‘x’ weeks
Figure 4.2 Measuring obsolete and excess.
4.C Obsolete and excess stock

Working sequentially down the Pareto stock curve will expose items that are either
obsolete or excess, so the inventory reduction should only consider these when
their value is reached on the Pareto curve.
Obsolete stock is defined as inventory with no foreseeable demand. For conve-
nience, this is often assessed by including items that have not moved for ‘x’ weeks:
‘x’ being appropriate for the industry. This is a useful definition, but has to include
other causes of zero requirement.
Excess inventory has usage but, at expected demand rate, the stock will last
a very long time. This is usually defined by stock cover greater than ‘y’ weeks
(Figure 4.2).
In general, obsolete and excess inventory is caused by changes in demand levels
coupled with poor forecasting and over supply. In detail, this can be
• inventory no longer required by the customer;

• components no longer used for the manufacturing process;
• excessive purchase quantity;
• item superseded;
• out-of-date inventory for items with finite lifetimes.
The basic strategy to avoid the problem is shown in Figure 4.3. For example, the
solutions can be analysed and future obsolescence avoided, as shown in Figure 4.4.
Having set in place the techniques for avoiding the re-occurrence of surplus
stock, there is the problem of getting rid of the current items.
4.C.1 Surplus stock valuation

Why is obsolete stock disposal important? The accountants put a provision in the
accounts, so this compensates for the value. However, this provision is taken
Improvement strategy
Outdated Excess Inventory held to Duplicated

products purchases avoid failures stockholding
Manage Improve Improve Manage

demand supply quality process
Improve data Co-ordinate

Increase delivery
accuracy, information
Improve frequency and
eliminate and
forecasting reduce lead
scrap supersession
time
avoidance process
Figure 4.3 Strategies avoiding surplus stock.

Based on Si Xiao Liu, MSc dissertation, 2015, WMG, University of Warwick, p51 (unpublished;
reproduced with permission).
Cause Solution Avoidance

Customer Over-production, poor Investigate any Collaboration, leaner
specific liaison with customer, resumption or manufacturing or supply,
products large batch manufac- alternative, use fair partnership contracts.
ture or bulk purchase enforce contract
discount. terms.
Standard Unreliable forecasting. Promotion, find Improve historical forecast
products Make to fill capacity. new customers, and market intelligence.
sell to discount Measure end user usage,
Unreliable supply. warehouse. lean supply chain.
Components Excessive purchase Find alternative Accurate master schedule,
(including quantity, ineffective uses. small batch quantities.
packaging dependent demand Return to supplier Closer customer liaison.
materials, etc.) control. or sell standard
items.
Figure 4.4 Avoiding surplus inventory.

directly from the profit, so the less obsolete and excess stock that is created, the
bigger the profit. It is important for the survival of the business. This directs how
this stock is dealt with.
4.C.2 How to reduce the surplus stock

The first step is to ensure that there are no outstanding supplies, and that the
system will not automatically order some more when the excess is disposed of.
Then the disposal project can commence.
The problem is to enthuse colleagues about obsolete stock – difficult, but just
concentrating on a few items helps. Taking just the six highest stock value obso-
lete or excess items to work on is advisable.
These items should be actioned in the sequence shown in Figure 4.5 and pro-
vided to the appropriate departments for a limited time (suggested in the figure).
All the techniques are not applicable to all items. One person should control this
process and have the authority to ensure it works. Where there is a good outcome
from the first six surplus items, the process can be repeated with the next six
(see also Chapter 7).
Sequence Department Technique Decision time

1 Marketing/ Sell items in new
Sales markets
6 Weeks
2 Marketing/ Promote to existing
Sales customers
3 Sales Discount price
4 Management Transfer to sister 2 weeks
companies
5 Purchasing Sell materials back to
3 weeks
supplier
6 Sales Sell as substitute for
inferior product
7 Technical Alter/repackage 4 weeks
8 Technical Reuse Major Modules
9 Technical Reclaim components
10 Sales Use to promote other 3 weeks
products (free add-on)
11 Purchasing? Segregate materials 4 weeks
and sell
12 Sales Donate to charity or 4 weeks
staff for promotional
purposes
13 Purchasing? Scrap and pay for 4 weeks
disposal
Figure 4.5 Sequence for dealing with surplus stock.

Key points
• Pareto Analysis of stock value is the basis for stock reduction.
• Reduce inventory of the highest value stock line first.
• Reducing supply batch size is the best method.
• Adopt a policy for avoiding the risk of obsolete stock.
• Attack surplus stock using the correct technique in a project.
Notes
1 ‘Stock line’ is an individual type of item (sometimes called an SKU).
2 The ‘minimum’ on some systems is set to the review level for triggering resupply even
though the stock continues to fall below this level until the delivery arrives. Here, the
‘minimum’ is the typical stock level immediately before delivery.
5 Management and control
• Inventory strategies.
• Identifying performance targets.
• Managing stock value.
• Skills, support and systems.
• How to value and depreciate inventory.
5.A Where stock control fits into the organisation

Inventory management is an increasingly important business activity, and is often
the differentiating factor in the success of a company. It is not enough for businesses
to advertise innovative items to the market – they have to be available reliably and
managed at minimum cost. This is the role of the inventory manager, although the
job may be designated ‘materials controller’, ‘planning manager’, ‘supply chain
director’ or similar.
It is not generally recognised that inventory management is a profession
requiring advanced techniques and a high level of expertise. Originally, Stores
personnel would advise Purchasing when stocks were low (Figure 5.1A), but
gradually there was a separation between the physical stores operations and the
management of inventory levels (Figure 5.1B). As lean operations developed,
there was then need to co-ordinate the supply and sales, and so integrated mate-
rials management developed (Figure 5.1C). The need for more supply chain
co-operation has led to structures where efficient flow of inventory is central to the
business (Figure 5.1D).
From a historically lowly role as part of stores people’s jobs, inventory manage-
ment has risen to a challenging, complex expertise for higher management; yet,
knowledgeable practitioners are scarce.
Inventory control is a senior management activity.

Management and control 63
A B Operations C Materials
Director
Director Director
Purchasing Inventory Inventory Customer

Purchasing Purchasing
Control Control Service
Stores Stock
Stores Stores
Control
D Supply Chain
Operations Director
Supply & Inventory Customer

Distribution Warehousing
Contracts Management Service
Figure 5.1 Evolution of supply chain operations.
Stock control encompasses:
• providing an appropriate level of stock for customers;

• maintaining minimum inventory to achieve this;
• organising supply (organising deliveries within contracts);
• forecasting demand;
• planning inventory against company targets;
• stock allocation and delivery promises;
• monitoring and controlling service and inventory;
• production planning and control (for manufacturing);
• continuous development of more effective and efficient operations.
These are organisational activities. The physical control of inventory, including

stores, despatch and movement, is a separate discipline, managed by a parallel
organisation which provides the basic data for inventory management. A similar
relationship can arise in manufacturing with the Production Control department,
who carry out the planning, and Production Supervision, which seeks to work to
the plans provided by the Production Control department.
Stores control takes care of inventory – inventory control sets the

amount to keep.
Inventory managers are reliant on the warehouse (or stores) not only to carry
out the physical receipt and despatch of the items, but also to provide information
to the rest of the company. It is particularly important for inventory management
that the information provided is accurate. Checking stock physically is very
time-consuming, and recording systems should be sufficiently accurate to make
this unnecessary apart from stock auditing.
Stock control in many organisations is carried out at a different location from the
warehousing. Thus, inventory controllers can manage on-site stocks or inventory in
many locations as long as they have correct information. Here real-time communi-
cations are a great benefit. This enables warehousing to be at the most economic
hub for transport and customers, while inventory control is located independently
at the most convenient place for management. Thus, centralisation of this expertise
benefits teamworking and has the potential to develop a centre of excellence.
As operations are increasingly complex, and targets getting tighter, there is a
greater need for more innovation and use of advanced technology. The technology
does not supplant inventory personnel; it takes the routine work away and supports
professional controllers in making more skilled decisions.
5.B Responsibilities and targets
5.B.1 Rationalising the problem

As there is often a myriad of different inventory lines, inventory controllers have
to take a simplified view of the stock and apply detailed control to large groups.
Every item has its own characteristics, often with a unique story: ‘Do you
remember when the supplier’s warehouse burned down?’; ‘The customer was
frozen up for five weeks.’; ‘It was put on a boat to Antarctica by mistake.’.
Originally, individuals managed stock through their memory and understanding.
Nowadays with tighter targets, reliable professional tools are required and exper-
tise is employed to enhance control of A class (generally a waste of time for C class).
In driving a car there is a distinction between the driver (who may not be able to
repair the car) and the mechanic (who may not be able to drive). In stock control,
most people are the drivers, but the stock control vehicle is not well designed;
mechanics are few, so the driver is often forced to make improvements. The driver
who understands the mechanics can always get better performance out of a vehicle
(this was also the case when cars were first invented).
The principles of all inventory management are the same – only

the situations differ.
The stock controller should largely ignore each item’s details and consider the
characteristics of demand patterns and supply. Despite the widely different types,
sizes, shapes, usages, origins and characteristics of the lines, whatever the item is,
inventory control remains the same. An arguable point? What about liquids,
explosives, perishable foodstuffs, pine trees? They are all very different. True, but
despite the differences, their supply and demand characteristics are not that differ-
ent: they depend on market conditions and not on the product. Just as financial
controls are applicable across different industries, so is inventory management.
The characteristics of the stock determine the mix of inventory techniques to use,
but the components are the same:
• demand forecasting;
• determining safety stocks;
• negotiating supply patterns;
• avoiding slow-moving stock.
Inventory techniques are common for all businesses and their skilful mixing for
aggregate and individual item inventory control allows many different types of
stock to be managed. Simplicity is the key, but this is simplicity of operations; the
systems need to be complex in order to give accurate control information. Simple
systems give superficial answers. Integrated systems (including enterprise
resource planning, ERP) often lack the detail required for decision support in
practice. Integrated systems enable communication, but it is up to the business to
take advantage of this. Usual pitfalls are:
• having different descriptions or units of measure for an item in different

departments;
• delay between the action and recording it;
• using processes where input errors are possible, so data is incorrect.
To achieve simple but accurate decision making, systems should include com-
plex tools such as software modules for forecasting, modelling, communicating,
but these need not concern the daily operation so long as they work correctly. The
actual logic of such tools has to be intricate to cope with the complicated nature of
inventory planning.
Firefighting is only necessary when there is a fire – planning avoids this.
Simplification should be considered as an ongoing challenge for development

of good stock keeping. Areas which can usually be improved are:
• accurate recording of stock in each location;

• real-time recording and systems;
• inventory management within an integrated business system;
• the same operating procedure applied throughout;
• paperless information systems;
• purchasing as part of inventory management.
The better these aspects are managed and staff are trained in inventory expertise,
the better the results. Professional stock controllers should not have to spend
their time chasing problems, but have time to work on inventory optimisation,
co-ordination and improving the level of control.
5.B.2 Key KPIs

Key performance indicators (KPIs) themselves vary in importance. As discussed in
Chapters 1 and 2, the aim is to provide the best availability with the least capital
and operating costs. For successful operations, KPIs should be set up and monitored
regularly (weekly or monthly) for each of these aspects. These KPIs shape company
policy, so the top management team should be involved in setting appropriate
values based on market and financial requirements and current performance
(Figure 5.2).
Aim for perfection: 100% availability, zero inventory, no operating costs.
Operational objectives Key performance Measurement technique

indicator for result
Customer Availability or delivery on time % achieved, arrears
Customer delight Survey, queries, complaints
Inventory level Total stock value Stock cover, gross value
Value of arrears Value past due
Operating costs Warehousing costs Actual vs budget
Distribution costs Supply and sales actual vs
budget
Improvements Projects undertaken Number and value started new
and ongoing
Profitability contribution Benefits identified (availability
or costs)
Obsolescence Value of stock written off Provisions, value disposed of
Necessary objectives Performance measurements

Develop staff Training days, budget, outcomes
Provide correct and effective information Investigation, queries, complaints
Meet health and safety requirements Accreditation, incident records
Meet industry standards and legislation Accreditation, awards, non-conformances
Surpass environmental and social Accreditation, awards, surveys
requirements
Figure 5.2 Management controls.

The KPIs should be chosen to meet the requirements for a successful busi-
ness, and at the same time look after its development of people, status and
conformance, and in the community. The process of setting targets usually
begins by measuring current success (satisfaction with current stock levels
is often due to complacency or not understanding that there are better
methods). Measurement is no use without action, so each record in Figure 5.2
should show:
• current value;
• target value;
• value of improvements currently achieved;
• plans for further improvements;
• actual current progress against plans.
The three dimensions that must be applied to each of the improvements are:
1 Quantity: how much the monitored values will change.

2 Date: the stages by which the quantities will be changed.
3 Responsibility: who will be accountable for achieving the results?
Good techniques enable companies to meet the short-term KPIs simultaneously,

and not to sacrifice one to meet another. It is then sensible to tighten the targets slightly
and work toward these. Measurement methods should be selected to meet the prime
business needs (in Chapter 2 alternative ways of measuring customer service are
discussed) and be supported by senior management.
A realistic view of KPI performance is essential.
When the full KPI results from the monitoring KPIs are available to inventory
controllers each week (or month), then improvements can be planned. Where the
information is available rapidly (say within three days), it can be used for improve-
ments (otherwise, commonly, it is used for avoiding blame).
Senior managers have to be aware that the art of correct management is to
set targets that meet company strategy, but which are both achievable and
challenging. They should reflect an expectation of performance, not an achieve-
ment under ideal conditions. This can be difficult when trying to co-ordinate plan-
ning with other departments. A good method of improving the accuracy of
planning is to give two targets, one optimistic and one pessimistic, and to ensure
that performance lies between the two. By adjusting the two limits, a target range
can be identified which is both attainable consistently and acceptable for planning
and co-ordination with other activities.
5.B.3 Other performance indicators
The KPIs are the reason that the inventory exists. Besides these, the company has
additional objectives to consider, including personnel and operational targets.
Customer service can be structured by setting availability targets for different
types of items. Some service is likely to be ex-stock with different availability
levels, while other service will be supplied on a range of lead times. Information
required to give fast direct performance measures include:
• number of customers and orders affected by shortages or late deliveries;

• analysis of shortage according to cause;
• frequency of supplying substitute items;
• details of customer care initiatives.
Set quantitative targets – or you will not know when you have achieved them.
Stock levels will differ for these product ranges and have different expectations
in terms of stock turns. Items with a ‘lean’ type of operation have very good stock
turns, while slow-moving items such as service parts can inevitably achieve only
relatively poor stock turn.
Besides these major KPIs, there are more requirements for the operations to
fulfil, since these prime external targets do not meet the legal and long-term needs
for the business.
Aspects which are equally important include quality management, credibility,
legal, motivational, health and safety, communication and process development.
The real requirements for inventory control are numerous and, where possible,
targets are needed to evaluate and monitor these objectives. Inventory manage-
ment must also meet general business objectives. As continuous improvement is
the only way to stay competitive, management of change and projects is part of the
job. Targets need to be set in the areas of:
• personnel development planning;

• improvements in systems and procedures;
• quality development for products, service and administration;
• maintenance of legal and quality accreditation;
• management of change processes;
• supplier audit;
• environmental issues;
• presentation, style and image of the department.
For well managed inventory control, all the requirements of the market, the
operating environment and the supply chain have to be considered. Greater under-
standing of the real objectives will improve the effectiveness of the business as a
whole and the status of inventory management within the business.
Focus on all the KPIs simultaneously.
5.B.4 Operating budget

Operating costs are increasingly the responsibility of individual inventory
controllers. Operational activities provide the basis for the costs, so the opportu-
nity for cost reduction can be reviewed. It is important that the allocation of cost
to the department only includes costs which are under their control. Monitoring
the following aspects can be fruitful in identifying potential cost savings:
• Value of transactions or number of transaction per day (e.g. customer orders

processed, stock reconciliations done, adjustments to stock levels, etc.).
• Processing delays or queue lengths (e.g. allocated customer orders).
• Cost per transaction by type (direct cost and allocation of overhead).
• Cyclic variations in workload hourly and weekly.
• Performance against budget for each activity.
By adopting a kaizen or Six Sigma approach, a succession of small improve-

ments can be made giving a gradual improvement in performance and enhanced
satisfaction for the employees.
5.B.5 Purchase commitment

One way of maintaining control over inventory value is to balance input and out-
put values. If the value of input is kept to below the value of output, stock reduces
(see Chapter 4). The value of supplier deliveries should be less than the forecast
value of customer despatches in each time period. Supplier deliveries are a result
of purchase order and schedule commitment, so it is the ordering process which
has to be kept under budgetary control. The control process relies on collecting
information on:
• the forecast value of demand for each time period (week or month);
• the amount of schedule or order commitment already made for delivery
in each time period;
• the typical value of last-minute (emergency) supply orders which are committed
each time period.
To reduce stock, keep value of receipts of issues each week.
A policy for the target value for stock reduction, or increase, has to be set for
each time period, taking into account what is considered practical or desirable.
From this information, it is easy to calculate:
1 Maximum delivery value for each time period
[= forecast customer demand – value of stock reduction target per period].
2 Maximum value of order commitment for delivery per period
[= maximum delivery book value – typical value of emergency orders].
3 Maximum value available for new orders placed for delivery during that
period
[= maximum value of order book commitment – amount of order book
commitment already made].
This maximum value available for new orders is the ceiling for total order value
for delivery in that time period. If it is exceeded by placing a further supply order,
that order should be rescheduled into a later time period (or another order delayed)
or else the resulting stock value will increase.
The current order book value with suppliers for delivery over the next 5 peri-
ods (weeks or months) is as shown in Figure 5.3. The anticipated value of
issues each period is shown in the first line. This is the maximum allowable
receipts into stores that period if the stock reduction programme of £12k is to
be achieved. Already the company has arranged schedules (£32k initially) and
other orders (initially £23k). The company is also expecting to need £19k
worth of last minute supplies to be delivered next period, but has plans to
reduce this to £8k in period 5. Therefore, receipts for next period are £55k
committed plus £19k anticipated. By keeping further receipts for the next
period below £14k, the stock reduction of £12k will be achieved. Figure 5.4
illustrates the control of inventory value using supplier delivery management.
Period 1 Period 2 Period 3 Period 4 Period 5
Expected issues value in £1000s 100 105 108 98 90
Receipts
Unallocated to spend Amount 14 29 25 28 38

Emergency orders Cumulative 86 76 83 70 52
expected
Amount 19 19 17 15 8
Stock reduction Cumulative 67 57 66 55 44
Amount 12 12 12 12 12
Delivery of items Cumulative 55 45 54 43 32
already ordered
Amount 23 14 22 18 6
Scheduled receipts Cumulative 32 32 32 25 25
Amount 32 32 32 25 25
Figure 5.3 Planning inventory value.

Expected issues Expected issues

value value
Unallocated spend
Emergency
orders expected
Stock reduction
Delivery of Items
already ordered
Scheduled receipts
Supplier receipts 1 2 3 4 5
week/month ahead
Figure 5.4 Controlling total spend.
5.C Skills and systems

The use of integrated systems within a business and for supply chains provides a
great deal of data, which has to be assessed, interpreted and, where necessary,
used. The inventory manager has to cope with two system problems:
1 data reliability;
2 application functionality.
Data reliability relies upon correct inputs from other users, and businesses
have to ensure that they apply failsafing (Pokayoke technique) to improve this.
Integrated systems often do not have the technical functionality for professional
stock control and so users are forced to use spreadsheets for detailed calculations.
Systems, besides performing transactions, enable the user to achieve the
objectives most conveniently. The term ‘user friendly’ is often applied, but seldom
true: the system should be designed to suit the type of person expected to use it,
their knowledge, skill and available time. It is important that each operator has
appropriate training to understand and control their activities.
The more professional user can cope with more complex system controls to
achieve enhanced results. A regular operator only requires an uncluttered transac-
tion screen with the system set up for fast input or enquiry, focusing on relevant
information. An infrequent user requires detailed instructions and explanation in
order to make effective use of the system. If they do not understand it, poor results
may follow. Although systems should be straightforward and user friendly, their
underlying logic should be complex enough to cope with the many variables that
occur in practice. For instance, a user with a seasonal demand pattern needs a
system with seasonal forecasting built in, otherwise it will calculate inflated
stockholding. Using seasonal forecasting systems (Base Series, Regression with
Dummy Variables, or Fourier Analysis) improves KPI achievement.
A car driver does not have to understand electronic ignition logic to be able
to drive. Similarly, the stock controller need not understand the details of the
forecasting tools as long as the answer is very good for the individual items and
there are appropriate controls for the operator. Two distinct features for choosing
a successful system are:
• apparent simplicity geared to user skill;

• system sophistication geared to the importance of the application.
Successful implementation develops the right balance between systems,

procedures and the user’s knowledge of a particular application. As an example,
take a computerised parts list. A customer requires an item, but does not know the
part number. The old hands in stock control with many years’ experience will just
quote the item number and look at the stock record, because they know it.
New starters require support in finding the number. How can this be achieved?
There are basically three methods, namely:
1 training;
2 support procedures;
3 system features.
and the different approaches are illustrated in Figure 5.5.
Training
• Applications and procedures training;
• Operation and reference manuals;
• Helpline and support.
System sophistication Professional expertise

• Usability; • Skilled inventory;
• System features built in; • Managers support;
• System does everything. • Specialists management.
Figure 5.5 Achieving better inventory control.

5.C.1 Training
If the user understands the application and the system, and is motivated, control
should be good. Training is essential for providing enthusiasm, understanding and
knowledge. Inventory personnel need specific job skills that enable them to gain
the full benefits of their systems and contribute to operating effectiveness.
Investment required for training is minuscule compared with the direct inventory
savings.
5.C.2 Support procedures

Where the user of the system is unlikely to have knowledge of the application (e.g.
technical knowledge by sales clerks), the company can provide the necessary
information through supporting procedures. In addition to the operating systems,
users – or even customers – can obtain more information for decision making.
Expert support can be given from centralised support or via online chat. Where
there are enquiries or issues outside the expertise of the operator, a similar
approach to call centres can be useful. A ‘decision tree’ can be set up by an expert.
This is a series of questions, the answer to one question leading logically to
another, which will be different depending on the answer. The questions in the
decision tree need to be yes/no or quantitative and are focused on providing a
satisfactory outcome.
For example, a person receives enquiries from customers needing advice on
which stock item to buy. This person must either gain sufficient technical knowl-
edge of the stock, pass the item on to an expert or have the information provided
for them by a system. The decision tree prompts a question, ideally with a yes/no
answer. This answer then directs the user to another question until the query is
resolved. This enables a person without detailed knowledge of a subject to be
effective. The expertise has been applied in creating the logic, instead of repeti-
tively while working.
Development of these procedures has to be overseen by a highly-skilled person.
Thus, there is an initial cost, but the procedure can then run and advice can be
given without the same level of expertise. This is a cost saving when the situation
occurs regularly.
5.C.3 System features

Inventory control data comes from the physical recording of warehouse move-
ments both locally and from remote stores and the supply chain. Successful con-
trol requires a high level of accuracy of this data. The purpose of the system is to
take this data and transform it into decision making information for the inventory
manager. Simplistic systems (e.g. where minimum stock levels are input by the
user) require a great deal of user expertise. Sophisticated systems provide
improved support, reduce routine work and give controllers time to respond to
market and other developments.
Historically, there were fixed ‘maximum’ and ‘minimum’ in systems, and these
were superseded by varying safety stock levels and delivery quantity concepts
built into the systems (see Chapters 8 and 11).
Make the system produce relevant information, not large lists.
Systems should analyse stock into ABC classes and use historical information to
forecast and calculate the (weighted) average (see Chapters 11 to 13). Once customer
service policy, supply lead time and demand variability have been established, the
system can calculate safety stock and supply quantities. The system can then work out:
• purchase schedules and orders;

• expediting requirements for suppliers;
Customer demand
Availability
This week/month Last week/month Target
Availability (%) Availability (%) Availability (%) Date
Backlog
Overdue More than Overdue More than Overdue More than

x days x days x days
Inventory
Inventory Slow Weeks Inventory Slow Weeks Inventory Weeks

value movers on value movers on value on
value hand value hand hand
Purchasing
Activity This week/month Last week/month Target
value value value % success

Orders placed
Goods received
Late deliveries
Emergency orders
No. of orders Qty Qty
Figure 5.6 KPI records for each stores.

• management control information;
• exceptions;
• unfulfilled customer orders.
It should also provide information so that the stock manager can work on:
• incorrect forecasts;
• wrong supplier lead times;
• excess and obsolete stock;
• specific situations (e.g. customer specials).
Good systems produce information which can be used without amendment

by the inventory controller. Some data may need modifying because the full
information was not available on the system, but this should be rare if the system
is designed properly. In practice, many systems fail to provide this full specification,
resulting in time being wasted on routine analysis and interpretation, time that is
much better employed on customer and supplier liaison or ensuring that the sys-
tem parameters are accurate. Therefore, it is important to use a system with
advanced stock control features, and to use it as the basis for managing all
inventory. A typical layout for reporting on performance is shown in Figure 5.6.
5.D Inventory valuation
5.D.1 Unit cost of stock lines

Accounting procedures for stock are usually organised by a company’s finance
functions. Stock valuation methods should not influence warehouse picking by
demanding individual items or batches to be picked. In almost all circumstances,
picking should be organised as a first-in-first-out (FIFO) process.
Do not associate costing methods with individual stock items.
As suppliers change their prices, the unit costs of products change with each
supply order. There are several ways of doing this, as follows.
5.D.1.1 Average value

A running average is the safest stock valuation method, and usually the preferred
option. The cost of new deliveries is added to the total stock value and the total
value spread over the new total stock.
For example, the stock of an item is 20 costing £10 each. Ten more are purchased,
which cost £13 each. The total value is now 20 × £10 + 10 × £13 = £330. The stock
is now 30, so the average cost is £330/30 = £11 each. Any demand reduces the
stock value by £11 per item.
Supply order Quantity Unit cost Variance Total
PO 1001 10 £26 £2 £20

PO 1002 15 £23 −£1 −£10
PO 1003 8 £28 £4 £32
PO 1004 15 £24 £0 £0
Standard cost £24 Total variance £42
Figure 5.7 Standard cost and variances.
5.D.1.2 Standard cost

Fixed item value is calculated financially and held the same for a long time,
usually a year, giving a consistent valuation. Differences from the standard cost
are considered as good or bad ‘variances’. Standard costs are used widely in
manufacturing, as they stabilise costs and overheads and provide better budgetary
control.
The standard cost set by the accountants is £24, but purchase costs are higher.
There is a purchase ‘variance’ when the supply orders shown in Figure 5.7 are
received. The operational objective is to beat the standard cost.
5.D.1.3 FIFO
Stock is valued at its purchase value. The oldest stock is assumed to be used first
(good inventory practice), so stock value is therefore the total of the most recent
purchases. FIFO is best used as an accounting conversion, but not for identifying
which stores items to pick. Stores stock rotation should be arranged through the
warehousing control system.
5.D.1.4 LIFO
Last-in-first-out (LIFO) issues are valued at the most recent purchase price,
leaving the remaining inventory valued at a previous (generally lower) value.
This minimises the profit on items being sold and minimises the value of the
remaining inventory. LIFO reduces profits reported by a company and decreases
total stock value. Again, it is only a valuation technique and not appropriate for
warehouse use.
For example, current stock of an item is 10, valued at £80 each. A new batch of
5 is bought at £120. Total stock value is now £1,400. If a customer buys two, the
stock will be reduced to 13. The last purchases cost £120 each, so the new stock
value for 13 items is £1,160. If the selling price is £150 each, the recorded gross
profit is £60 (using FIFO, the remaining stock of 13 is valued at £1,240 and the
gross profit is £140).
5.D.1.5 Replacement value
Stock is valued at the current purchase price for replenishments. Using the
replacement value (e.g. sell at purchase price plus 100%) inflates inventory value
and requires market costs to be known.
5.D.1.6 SKU
The stock-keeping unit (SKU) is especially useful in distributed warehousing.
SKUs take into account that stock value depends on where the item is held. Say a
customer requires an item, which can be supplied either from the main stores or
from a satellite store and the delivery cost is £15 from either source. Which store
should deliver the item? Since stock in the satellite store has already been trans-
ported from the main store (probably also costing £15) it would be cheaper to
supply from the main store. Satellite stock has already cost more.
SKUs are ‘stock at a specified location taking into account the transport and
purchase costs’. If there are warehouses in two different locations containing the
same 3,000 lines each, then there are 6,000 SKUs because of the distribution
costs. The logistic cost has to be added to the stock value of lines (measured as
above) for SKUs.
SKUs take into account logistic costs.
These alternative valuation methods are illustrated by an example in Figure 5.8.

The ‘cost of the next item to be issued’ is the unit value of the next stock item
which will be issued to fulfil customer demand. Each of the valuations gives
different unit costs. For LIFO and FIFO methods, allocation of the costs depends
on assessing from which supplier delivery the next sales item should be taken.
5.D.2 Aggregate stock valuation

Stores stock can be valued either at full value or by ‘written down value’ to
acknowledge the fact that some stock is no longer usable or saleable. There must
be an amount of money deducted from the stock value to account for this.
The company has to acknowledge that inventory is unbalanced as a result of
mistakes by controllers, purchasers or marketers. They have squandered company
profit on useless stock. To reduce this, managers need to improve forecasting
departmental co-ordination, KPIs and control.
Writing down the value of non-moving stock is unfortunate, but essential.
The accounting code of practice requires stock to be valued at ‘the lesser of its
purchase cost and realisable value’. This means that items bought at a discount
have to be valued at a discount, and items with no likely use are valued at zero no
matter what they cost.
Date (week) Transaction Stock balance Cost of next item to be issued
(quantity)
Type Quantity Unit cost FIFO (a) LIFO (b) Standard Average Replacement
cost (d) cost (e) value (c)
*Stock check 10 20 10 20 20 24 20 20
1 Customer 3 7 20 20 24 20 20
demand
Supply order 1 10 26 26
3 Customer 2 5 20 20 24 20 26
demand
Supplier delivery 10 26 15 20 26 24 24 26
4 Customer 4 11 20 26 24 24 26
demand
5 Customer 1 10 20 26 24 24
demand
Supply order 2 10 28 28
6 Customer 6 4 26 26 24 24 28
demand
7 Customer 3 1 26 26 24 24 28
demand
Supplier delivery 10 28 11 26 28 24 27.6 28
9 Customer 4 7 26 28 24 27.6 28
demand
Supply order 3 10 24
10 Customer 2 5 28 28 24 27.6 24
demand
Figure 5.8 Alternative methods of determining unit stock value.

*
Existing stock of 10 is valued at £20 each.
The two approaches to writing down value are:
1 Depreciation: The risk of never selling an item increases the longer it is in

stock. Depreciation is a way of taking this into account. The rate at which this
happens will depend on the item: fresh food loses value very fast; aircraft
spares depreciate very slowly. The simplest and most common way of calcu-
lating depreciation is ‘straight line’ – deducting a portion of the item value
each year it remains in stock. Taking 25% off the original value each year
means that items have no value after four years.
Depreciation steals profit to cover for mistakes.
Another way is to take off 25% of the current value, so that the item always
retains some residual worth.
2 Provisions: A second way of reducing stock value is to make a ‘provision’ for

the total value likely to be lost in a year. It can be based on a detailed
depreciation calculation or a simple assessment. Whereas depreciation is
applied to individual items, provisions are applied to the total inventory. This
enables the provisions ‘reserve’ to be used for whatever incidents cause stock
obsolescence (sudden changes in legislation, contract loss, etc.). Provisions
are financial judgements. They can therefore be used in situations where there
are impending changes that affect inventory, but which are not predictable
from previous information.
Professional stock controllers do not use write-down as a method of reducing

stock value, because this only reduces company profits to pay for the write-down
(the proper methods for reducing inventory value were discussed in Chapter 4).
Good inventory management practice values stock at full value and applies
depreciation or provisions later, just for the financial accounts. This ensures that
slow-moving, high value stock receives some management priority.
Key points
• Identify the responsibility for stock management.
• Set quantitative KPIs and improve continuously.
• Value all inventory.
• Foster professionalism.
• Design systems and information for the user.
6 Lean supply
• Lean operations to increase profits.

• How to become lean.
• Just-in-time, the ultimate inventory system.
• The just-in-time supply chain.
• Inventory and distribution costs.
• Quality management.
6.A Lean supply philosophy

Inventory exists because items have been bought before they are required. It is
normally uncertainty or over-caution that causes stock. Best practice in inventory
management provides excellent availability for customers without significant
stockholding. This is lean supply: have items when they are needed and none
when they are not needed.
The idea may be simple, but the application of lean supply has given businesses
the opportunity to decimate stockholding without affecting customers. Instead of
trading availability and stockholding, as discussed in Chapter 2, the trade-off is
between co-ordination and stockholding. Better organised and controlled supply
chains require less inventory.
Lean supply is the result of high-quality co-ordinated supply.
Companies which are considering how lean operations can benefit their business
should realise that it is a combination of high quality and reliability. Lean thinking
provides a company with:
• minimal investment in inventory (improved cash flow);

• flexibility to respond to customer requirements;
• simplified controls for the operations;
• more involved and motivated staff.
Lean supply 81
Conventional Just-in-time
Satisfied with the status quo Continuous improvement
Lead time is fixed Reducing lead time is a continuing challenge
Product range is a sales issue Product range reduction is an inventory issue
Management provide methods Operators are responsible for practices
Stock in case of customer demand Purchased to meet demand rate
Convenient purchase batch size Buy singly or small quantities
Figure 6.1 Contrasts between conventional and JIT inventory control.
Stock cover Inventory investment (£)

52 weeks 2,500
10.4 weeks 500
6 weeks 288
4.2 weeks 202
1 week 48
1 day 10
1 hour 1.3
Figure 6.2 Warehouse stockholding.
The involvement results from some fundamental differences between lean

philosophy and conventional thinking. Systems are set up to make current
operations work effectively. However, we know that things could be improved.
Lean thinking is that continuous improvement is an essential part of operations.
If the business is not making changes, then it is falling behind the competition –
because they are making changes. Some contrasting features of managing stock in
a conventional and in a lean manner are shown in Figure 6.1.
In Chapter 1 two companies were described, each of them had a £5 million
turnover. Like many businesses, half their costs are in purchased materials.
Lean philosophy says, ‘don’t be complacent – there are always

ways to improve’.
6.A.1 Warehousing example

If these companies are distributors, then the stock cover calculation for Slack &
Co. is 52 and for M. Tight Ltd is 10.4. Applying a lean approach to selected lines
(product types) could reduce the overall stock cover to one week, and would result
in an inventory holding of only £48,000 (a saving of £202,000 even for M. Tight
Ltd). The stock cover and relative inventory investment are shown in Figure 6.2
for a turnover of £5 million.
Lean operations not only reduce the inventory value, but also mean less inven-
tory to count and manage. It decreases the complexity of control and can reduce
82 Lean supply
the size and cost of warehousing. Of course, lower stocks expose the faults in the
physical and information processes and these have to be improved before further
reductions can be made.
6.A.2 Manufacturing example

Applying the same logic to a manufacturing situation shows additional benefits
through the savings in planning resource. The situation is more complex in manu-
facturing. The material cost starts out as, say, 50% of turnover, but processing then
adds value until it is worth typically 70% of sales value. For simplicity, let us
assume that there are equal amounts of inventory throughout manufacture, so the
average inventory value is 60% of sales value.
Typically, a company with a turnover of £5 million issues about 50 new jobs to
production each week (the logic of the following discussion works just as well for
companies with different numbers of jobs and turnover values).
Manufacturing lead time results from poor product flow co-ordination.
Using these figures, a manufacturer’s performance results are shown in Figure 6.3.
Stock cover in manufacturing is required for the duration of average throughput
time. For Slack & Co., the £2.5 million inventory equates to 43.3 weeks’ cover and
a total of 2,167 jobs in work – assuming that all the inventory value is in work-in-
progress (WIP). For M. Tight Ltd, the WIP equates to 8.7 weeks’ cover and 433 jobs.
The advantage of reduced lead times in manufacturing is twofold, as can be
seen from Figure 6.3. Inventory investment is reduced as well as the number of
jobs and the complexity of the materials management. It takes a great deal of
effort to control 433 jobs, but much less to track 50.
The illustration shows that major cost savings go hand in hand with reductions
in manufacturing complexity. Inventory cost savings can be achieved by cutting
the lead time down to a few weeks, days or hours – and then benefits flow from
other operating costs, such as less shop floor space required, improved flexibility,
less expediting, more control and less material handling.
Throughput time (weeks) Inventory investment (£) Number of jobs in progress
43.3 weeks 2,500 2,167

8.7 weeks 416 433
6 weeks 288 300
4.2 weeks 202 210
1 week 48 50
1 day 10 10
1 hour 1.3 1.35
Figure 6.3 Effect of throughput time on manufacturing inventory.

Lean supply 83
Furthermore, the people who can see which improvements to make are those
who are working the process. Instead of operators simply carrying out the tasks,
they should have the opportunity and time to discuss and develop the processes.
The bases of successful lean operations are encapsulated in the nine principles:
1 Customer focus: Identify what the customer really wants and exceed their
expectations. This means providing a solution to the customer’s situation
though good relations, flexibility and agility.
2 Value and value-added: Define ‘value’ in terms of what the customer needs
and wants. Only do activities that enhance customer value.
3 Waste: Any material, process or action that does not contribute to customer
value is waste.
4 Flow: Provide goods to the customer at their rate of use. Make all processes
within the business work at this rate. This entails small batch supply, consis-
tent variety, uniform throughput volume, reliability and, for production
processes, quick changeovers between products and cellular manufacture.
5 Pull: Instead of providing items to a forecast (or MRP)1 on long lead times,
trigger demand by consumption and have a supply arrangement so that
resupply is fulfilled within the day (or within the hour if the replenishment is
internal). The mechanism for this is ‘kanban’ control.
Use one, get one and don’t hold stock.
6 Quality: Lean supply only works successfully where there is consistent high
quality. This can be achieved by building in quality at each process rather than
inspecting failures out. This requires careful design of tasks, products and
processes including the reliability of equipment, which can be improved
through ‘total productive maintenance’.
7 Supply chain: A well-established, committed supply chain enables consis-
tency to be maintained. If several suppliers are in the same ‘supplier park’,
then the logistics are simpler, small load economics are improved and perfor-
mance is enhanced.
8 Operator involvement: Many problems in supply and storage are well known
to the people carrying out the tasks. Effectiveness in supply and administra-
tive process can often be improved greatly through small changes arranged by
the people carrying out the processes. If it is in their interest and responsibil-
ity to make agreed changes, then major improvement can be achieved without
significant capital expenditure.
9 Continuous development: The means of making changes should be through a
series of small projects carried out by these operators, who are allowed time
for this. The tasks are identified by these personnel and the job of management
is to encourage, provide external resource if needed and to champion the
outcomes, timescales and resources used.
84 Lean supply
The cumulative effect of these improvements makes stock superfluous, and there
can be a gradual reduction toward ‘just-in-time’ supply (JIT). It is not possible
to ‘introduce JIT’, but JIT will result from the optimisation of all these aspects
(1 to 9). The techniques that will lead to JIT include:
• Supply what is required in the quality required.

• Get deliveries when required – not too early.
• Plan with the supplier so that the effective lead time is less than a day.
• Ensure high quality of records, supply, delivery, forecasting and target setting.
• Use all the expertise available (i.e. operator knowledge plus technical back-up).
• Arrange small, ongoing development projects for shop floor personnel.
These are the fundamental changes which lead to JIT – all good inventory manage-
ment techniques.
Inventory avoids synchronising supply and demand.
6.B Lean principles
6.B.1 Lean focus

Global competition and customer pressure has made it imperative that businesses
apply lean principles to survive. However, in practice it is often challenging to apply
lean operations throughout. The approach should be to review the nine principles
described above, then find the area that is a significant issue and could be solvable.
Business needs to use JIT to remain competitive.
Getting teams working on a single aspect first, and improving that situation,
gives confidence and makes people realise that change is beneficial and inevitable.
An attitude of ‘we’re doing alright – don’t risk changing anything!’ has to be
replaced by ‘The commercial world is changing, and we’ve got to be ahead of the
game; now, how do we make it happen?’. How are requirements changing?
• Customers want things faster – shorter lead times or ex-stock.

• They will pay less for them and require additional services.
• They want to receive them in smaller quantities on a JIT basis.
• They want higher quality, higher specification, more diversity.
• Everyone can get facts and rumours via social media.
• Competition is from low-wage low-cost economies, so profits are squeezed.
• Regulation, health & safety and a suing mentality are increasing operating
costs.
Lean supply 85
Businesses are responding by offering solutions rather than products, outsourc-
ing, market specialisation, supply chain co-operation and cost reduction. Lean
developments are a major contributor to cost reduction. Employees have a stark
choice: become more efficient and probably lose some jobs, or remain the same,
become uncompetitive and lose all the jobs. If effectiveness is improved sig-
nificantly, market share is increased and more jobs could be created.
Developing lean operations requires:
• A culture of change. Kaizen developments, where small teams of operational

people carry out practical projects to improve effectiveness and their own
satisfaction (not requiring significant capital cost). Alternatively, an approach
such as Six Sigma, using a professional hierarchy, can create the changes.
• Customer orientation. Success results from providing the correct combi-
nation of availability, customer relationship management, value and product.
• Reliability. Unreliable supply (externally or internally) requires extra
inventory to mitigate the risk. Reliable supply enables inventory to be reduced,
even when lead times are long.
• Quality. The philosophy of Pokayoke (failsafing) is an essential part of
making operations lean. Starting from Murphy’s law – ‘if things can go
wrong, they will go wrong’ – companies need to accept this and counteract it.
Improved design of both products and processes ensures that supply is
reliable. If it is impossible to make a mistake, then it will not happen. There-
fore, poor quality is the manager’s fault, not the operator’s. The manager’s job
is to minimise the quality risk throughout, including product quality, delivery
and all types of data in information systems.
• Planning. Accurate forecasting and planned supply enables goods to flow
along the supply chain without panic or disruption.
These are the factors that will lead to a reduction in stockholding and ultimately
provide JIT operations.
Performance is dependent upon a number of factors – for example, suppliers,
the market, the company itself and the individuals within it – and these factors will
always throw up a problem of some sort (these should be considered as the
challenges of everyday stock control).
In high-volume logistics and manufacture, JIT supply is the natural way to
work: lead times and cycle times can be kept very low. In other operations, where
the demand is variable and uncertain, then the application of lean has to be
arranged carefully to be effective. The ideal situation for a company to be lean is
shown in Figure 6.4.
A major element not included in this ideal list is ‘personal commitment’. Lean
supply has been applied successfully since last century in many industries, notably
in automotive assembly where all supplies are JIT and supply failure does not
happen. Personal commitment is a major factor. Successful lean JIT operations
depend on commitment by people at operational level and equal dedication of
managers and directors. Many managers struggle to adopt lean philosophy and find
86 Lean supply
Attribute Distributor Manufacturer
Product range Mix and match products Only one process route
High volume Consistent demand Continuous, not batch,
processes
Stable market Good forecasting Accurate master schedule
Influential position Many customers Regular daily demand
Product quality Product assured by supplier Built-in process quality
Quality management Consistant and accurate Accreditation and Pokayoke
information
Local, reliable suppliers Daily pull supply Use-one-get-one approach
Dependent suppliers Significant customer for Controls supply chain
suppliers
Fast cycle processes Fast efficient despatch Synchronises process speeds
Figure 6.4 Ideal situation for just-in-time operations.
excuses for retaining ‘batch and queue’ systems, especially in an international

supply chain. The usual arguments are based on assumptions about distribution
costs or using conventional overhead costing rather than real cash flow accounting.
6.B.2 Repetitive processes

Optimising lean supply works best in repetitive processes, whether in warehous-
ing or in manufacturing. Many business sector customers require the same or
similar products, so this is not really a major constraint. For a usage of 20 per
month, a traditional approach could be to buy 20 at a time, whereas lean JIT would
require one per day. Such a small order rate may not appear economic for one
item, but it would be when dealing with a large variety of items.
Consider a production or warehouse packing line that each week completes:
• 2,000 of product A11;

• 3,000 of product B22;
• 4,000 of product C33;
• 2,500 of product D44;
• 1,000 of product E55.
Instead of working on a complete batch of each product, it could be better to

provide each item at the rate at which they are required. The overall lean plan can
be to provide at a balanced hourly rate (Figure 6.5). However, as customer demand
may vary, this can be adjusted daily using a kanban pull system. In practice, a
buffer stock is kept somewhere in the system unless the demand rate is very stable
(e.g. another process). Generally, fast-cycle JIT supply works this way internally,
since it balances capacity. The safety stock elsewhere in the supply chain copes
with any batching and variability of customer demand.
Lean supply 87
Item Throughput per week Throughput per day Cycle time (secs)
A11 2,000 400 68

B22 3,000 600 45
C33 4,000 800 34
D44 2,500 500 54
E55 1,000 200 135
Total 12,500 2,500 11
Figure 6.5 Continuous supply cycle times.
Arranging the same output each day makes capacity and supply
planning effective.
Examples of this situation are the automotive industry and high technology
equipment manufacture where supply is geared to a constant output rate and
resources organised for efficient and cost-effective supply (and production also).
In Figure 6.5 the activity completes an item every 11 seconds. This could be one
process line so that the throughput is not sensitive to the product mix.
Example
Automotive assembly generally has one production line, with operators
at workstations each carrying out a specific process on a different
vehicle at the same time. In a mail order warehouse, the capital
investment per line is not so great, and items are more diverse, so that
generally there is a number of specialised packing lines.
Inventory managers apply lean techniques to the most important items where it
is easiest to use. These are the ‘A class’ items with high usage, more consistent
demand patterns and significant financial benefits from inventory reduction. These
products will be controlled through JIT while the rest are managed conventionally
(see Chapter 8). High usage and turnover value items are therefore the focus for
lean and gradually a hierarchy can be developed for the remaining items in the
supply chain (see section 15.C.3).
Using a kanban pull system, described in section 6.B.3.1, means that workload
fluctuates with customer demand, which can be a problem if there are big
variations, especially as demand needs fulfilling immediately. Some demand
smoothing through order management or stock can be used.
88 Lean supply
6.B.3 Just-in-time supply
To minimise inventory, goods are despatched as soon as they are received. The trick
is to ensure that items are supplied in the same quantity and exactly when they are
required. This means throwing out the old idea that batches are beneficial and
replacing it with variety supply, flexibility and ‘flow’. Lean supply timescales are
hours rather than days, supported by longer term plans. The aim is to provide the
goods at the rate that the end user in the supply chain is consuming them. Where
suppliers provide a range of items and achieve high quality and delivery perform-
ance, a JIT system is obviously appropriate. Single source supply (see section 9.C)
assists in achieving this throughout the supply chain: the supplier should be treated
as an integral part of the organisation.
There is a balance between local lean sourcing and global long lead time, low
unit cost sourcing. As worldwide sourcing incurs extra costs for delivery,
administration and inventory, the situation is not straightforward. Many stock
controllers consider that receiving large batches reduces the workload and risk of
stock-outs without considering the effects on inventory value and obsolescence
(a well-known Spanish fashion retailer has half the obsolescence of its rivals
because it has a JIT supply chain). As products become more short-lived and
variety expands, smaller batch sizes are more suitable and profitable.
The lean ‘pull’ approach only gets supply after despatching the previous one,
whereas conventional stock control and material requirements planning are essen-
tially push systems. In the push system, the stock is provided for the next stage of
supply, e.g. buying items for sales to sell, or starting manufacture without having
the total production path clear. The philosophy with push systems could be thought
of as pushing so much into the warehouse or production plant that some items will
be forced out because there is only limited space!
With a pull system, the sequence is triggered by the consumer using an item from
stock. As soon as this is recorded, a replacement item is supplied, either from outside
or from a production process. It is organised so that the effective supply lead time is
very short (under one day) and the quantity supplied is small – a few minutes up to
a day’s worth of demand. In production, the demand cascades backwards through
the processes, creating a demand down the bill of materials and ending up with issue
of one product’s worth of raw material. This then initiates the external pull from
suppliers on a daily or (where there is international supply) weekly basis.
As any costs that suppliers incur are included in the sales price, it is often
advantageous from a risk standpoint for customers to collect goods from suppli-
ers, and to incentivise (financially) vendors to have the items ready.
6.B.3.1 Kanban
The way to operate a pull system is with kanban. It is simple and does not require
media. It consists simply of a ticket, or sometimes two. When an item is used,
the request for a replacement must be transferred back to the source to provide
another. This information can be a card kanban or a simple signalling system.
Lean supply 89
The card will identify the item, the quantity required (ideally one, but possibly
more) and where it is required. It will also state times so that performance can be
monitored. This ticket or ‘traveller’ has to be provided to the source very rapidly
so that the supply can be prompt.
The ticket is held with the physical goods and is an easily maintained and sim-
ple method of informing the previous stage of supply. Other options, such as
coloured lights or sound, can be used internally and, for suppliers, IT that uses
comms, apps or even emails could suffice. As long as the process is instantaneous,
completely reliable and accurate, any process is suitable.
Once manufacturers have solved consistency, quality and breakdown issues,
kanban supply is obviously the best way to organise for efficiency, agility and least
inventory. A flow line can have single item in-line stocks or very small batches
between each process stage. The trigger for manufacture is simply ‘when one is
used make another’ starting from the last process and feeding back gradually to
the first. Restructuring production into flowlines throughout a factory is sometimes
difficult because products have different process routes and there are bottleneck
processes. Organisation of production into a cellular structure enables kanban
processes to work within the cells.
For purchased items, the supply is triggered by consumption. The delivery
batch would be expected to meet the demands for a few hours or up to a day.
As the lead time is normally over an hour, supplier kanban size will be designed
to ensure that the recipient does not run out before the delivery arrives.
It is not possible to use pull processes with long lead time supply (a week or
more), and conventional inventory control is required. The kanban system has
the same basic process as the 2-bin system, with which it is often confused. The
kanban system is predominantly for A class items with fast cycle supply, whereas
the 2-bin system is for fast-moving C class items with large batch supply and
sometimes long lead times.
Lean JIT is established as the prime way of organising the supply chain and
manufacturing. It has decimated inventory and given a competitive advantage for
those who are continuing to develop it. Companies have individuals always
working to improve product quality, process efficiency, information systems and
operating value-added activities while eliminating non-value-added activities.
JIT results from best practice in inventory control.
6.C Implementing lean

The lean supply chain may be an aspiration, but it is difficult to implement in
practice. Creating the right conditions has to be worked on throughout the organ-
isation and in partnership with suppliers and customers – it does not happen by
accident. Lean supply should be considered as a quality process. Many failures to
implement lean supply result from poor control of quality (in data or forecasting),
90 Lean supply
complacency or lack of understanding (or usually both), resulting in incorrect
stock levels. There is no perfect solution to stockholding, but, like any other
quality improvement process, a lean approach gradually changes an existing
unsatisfactory situation into an improved one. Improved reliability and simplifica-
tion of processes should be the aim of all inventory managers. The more time
saved the better, and continuous improvement means reducing the timescales.
Time is a value-added commodity and wasting it is unprofitable.
The basic requirements to succeed in reducing timescales are:
• desire to improve;
• simplification;
• demand-led supply (pull);
• quality conformance;
• reliability of supply;
• devolution of responsibility.
These concepts are fundamental to good inventory management anyway

and require good communications and management, driven by the need for better
service and lower costs.
6.C.1 Avoiding stockholding

The impetus for lean supply comes from avoiding warehouse space. It is the
obvious way to organise the fresh food supply chain and other high-volume goods
where delays could cause high stock value. The question is: how can we supply
customers with a minimum of inventory? The expansion of internet sales at the
expense of the high street means that the internet supplier needs one stock location
or immediate production facility, whereas a high street chain has high multi-
location stockholding.
6.C.2 JIT environment
6.C.2.1 JIT and distribution

JIT requires more frequent deliveries from suppliers, but this does not automatic-
ally lead to higher transport costs, although it may require a rethink of the arrange-
ment for supply. Situations where there are no extra transport costs are where:
• freight is charged simply by weight;

• shipments include more variety of items instead of large batches;
• freight is charged at a flat contract rate.
Lean supply 91
Costs might increase where:
• the same delivery method is used more frequently;

• different suppliers supply each product;
• a flat rate per consignment is charged;
• transport is arranged one journey at a time.
So, these situations are to be avoided.
Change transport method and get more frequent supply.
Lean supply requires close co-operation with logistic suppliers as well as

product suppliers. By moving toward longer term, higher volume commitments
with suppliers, prices can be reduced based on the total traffic volume over an
extended period, if it is significant. The distributor requires a continuity of demand
and the opportunity to optimise routes, loads and schedules well ahead and this
can reduce operating costs for the distribution partner.
People are sceptical that it is possible to increase deliveries without increasing
the overall price. Others who have actually tried increasing delivery frequencies
have found distribution partners who can provide the service without increasing
financial disadvantage, since it gives them a regular base load.
The general cost savings produced by lean supply are discussed below. The eco-
nomics of frequent deliveries can be added to these savings for delivery processes.
Some of the typical benefits are:
1 Smaller delivery quantity means smaller loads, which use different transport
methods (carrier, shared loads, milk round), which often cost considerably
less since the operating costs are reduced.
2 Frequent fixed route deliveries enable a carrier to have a base load upon which
to build other business.
3 Regular distribution requires little management effort, once set up.
4 Discounts on contracts can be very significant.
5 Standard pack quantities and containers are smaller, less expensive and
recycle faster.
Lean distribution saves costs if organised properly.
The secret of success in saving costs on frequent deliveries is to:
• plan carefully so that delivery schedules are not altered at the last minute (this
is easier with JIT since the planning horizon is relatively short);
92 Lean supply
• be bold – radical changes in delivery methods are required if costs are to be
minimised (this often means a complete change in transport method);
• negotiate carefully – transport costs are adjustable because they contain a
large fixed element of cost; the apportionment of this cost (e.g. running lorries
from A to B) can be carried out in any way and transport suppliers may be
persuaded to apportion the costs away from routine major customers;
• be open – share the size, frequency and costs of delivery required with the trans-
porter, and knowing their relevant costs should ensure that their service is viable.
6.C.2.2 Quality management

Inventory management would be simple if quality was good; not just the products,
but particularly the stock, demand forecast and suppliers. Quality has been a focus
for many companies and many standards have been set up to promote good results
and effective operations.
In addition, the development of the total quality management (TQM) philoso-
phy of eliminating waste and improving operating effectiveness is the same as lean
philosophy. Therefore, TQM has been able to lead improvements that are required
for lean to work. TQM has focused management attention on productivity
improvement, standardisation of procedures and packaging, improvement of the
quality of delivery on time, accurate record information, operating effectiveness
and the complete operating processes of the company.
Companies that cause quality problems should foot the bill.
Quality management principles integrate well into stock control: it is a basic

assumption of inventory systems that the items supplied are fit for purpose. If a
delivery batch is not acceptable, then it has to be replaced immediately for the
stock control calculations to be correct. Delivery in smaller batch quantities often
reduces the overall effect of rejected supply since it impacts on fewer customers’
orders and leaves less time before the next delivery. Non-conformances cause
extra work, especially:
• creation and holding of extra records as a cross check;

• quality inspection on goods inward (also causing booking delays);
• sales personnel requiring physical stock checks rather than relying on the
recorded quantities;
• time spent investigating and checking;
• arranging replacements and priority deliveries.
Accurate inventory systems are both a quality and a stock management

concern. Improving efficiency in information systems is a continuing challenge.
Lean supply 93
Non-essential activities can be phased out, particularly repetitive tasks and
responding to inadequacies of the operating systems and errors. People will com-
plain continuously about a problem, but do not feel empowered or motivated to
solve it. Office procedures are a target for continuous improvement just as much
as shop floor activities.
The focus for all these activities is to enhance the customers’ experience, which
will lead to enhanced long-term profit.
6.D Operational benefits of lean supply

How can it be more efficient to deliver and manufacture in small quantities,
and increase the amount of administration? This is the traditional viewpoint.
Efficiency comes from bulk production and logistics. Equally, someone who only
knows lean JIT would think:
• warehousing does not add value, therefore eliminate it;

• do not store stuff because others are unreliable;
• the stockholder must have a lot of cash to spare;
• why buy ten when you need one? You will need to store them and probably
throw some away eventually.
Most companies have a large investment in inventory. Earlier in this chapter the
two companies from Chapter 1 were discussed, and the increase in profits from
leaner stock. If M. Tight Ltd could introduce full JIT in supply and demand, this
could bring the stockholding down even further to one or two days’ worth, say
£20,000 total stock value. This would release a further £480,000 cash for the
business and increase the return on assets from 16.7% to 20% as a minimum
(changing to JIT also eliminates the stores and some overheads, so the real
improvement is greater than calculated here). These companies have invested in
fixed assets, but, where most assets are rented, the percentage increases in profit-
ability are much larger.
The operational benefits arising from lean JIT are:
• reduced capital investment;

• ‘supply-to-order’ instead of ‘provision for stock’ (see section 8.B);
• shorter forecasting, so less slow-moving stock;
• better flexibility;
• simplified administration;
• eliminated waste processes;
• less scrap produced.
Each of these gives long-term cost benefits that are much greater than the one-
off costs of the reorganisation or additional workload. This could arise because
other systems are not modified. For instance, monthly delivery batches can be
ordered, receipted and invoiced individually. When this is changed to daily
94 Lean supply
delivery, the administration system would be overloaded if it were not now
changed. Purchase orders become schedules, invoices become statements, goods
receiving systems are simplified (no inspection) and moved to user areas. This is
often a gradual change, with some hiccups on the way.
6.D.1 Process flexibility

Suppliers give discounts on bulk purchase, which infers that it is in their interest
to sell large quantities. In fact, it is more beneficial to supply regular significant
quantities because it improves forecast accuracy and reduces stockholding. The
challenge is to provide variety items at the rate of usage throughout the supply
chain (Figure 6.5).
Many manufacturers see bulk manufacture as most efficient because they have
not embraced this lean concept. Many processes are incapable of changing product
instantaneously or making small quantities efficiently. Rectifying this is a priority,
especially as customers require greater variety.
Example
If changeover for a particular item type takes 10 minutes, conventional
logic says there is no point in making just one (which takes 15 seconds);
better to make at least 80, to be efficient. Accountants usually support
this type of simplistic logic because of their inappropriate apportioning
of overheads. This results in large batches being produced, fluctuating
workloads and high stocks (this cost is an overhead and therefore not
considered). The real situation is that changeover time is a variable
depending on what the change is from and to. If the process is not
required for anything else, it is better to spend the 10 minutes making a
changeover, even if the average stock value turns out to be weeks’
worth rather than days’ worth.
Fast changeovers give the opportunity for lower stockholding. On a packing

line, the change from one type of item can be fast if they are packed in the same
type of box, but slow if the box size has to be changed. The secret of success is
therefore to always run the right job sequence so that the number of short change-
overs will be large and long changeovers small. This requires planning and load
balancing, which is difficult when forecasting for weeks ahead, but easier when
looking ahead a few hours, as required by JIT processes.
If the items in Figure 6.5 are all processed on the same line, then one item is
required every 11 seconds (5.6 per minute). If the process is carried out on a cyclic
basis, as shown in Figure 6.6, then the cycle is repeated every 4.5 minutes. The
quantities, output rates for each cycle, are illustrated for an average output rate.
Lean supply 95
Repetitive cyclic job sequencing reduces overheads.
To cope with the variety in the 4.5-minute cycle, the changeover times between
the items need to be minimal. In general, changeover time should be reduced by:
• planning – effective job sequencing. By studying the actual time between a

pair of jobs, the total amount of changeover time can be minimised (the first
step is to avoid any emergency jobs that disrupt the sequence).
• operations – reducing changeover times. More fundamentally, operators and
process designers should be pressured into working tirelessly and continu-
ously to make changeovers extremely fast.
Changing the production philosophy from weekly batch quantities (Figure 6.5)
to cyclic lean JIT (Figure 6.6) will reduce stock levels by a factor of 20, and also
save overall cost.
6.D.2 Lean economics

Lean businesses have fewer errors, panics and changes in plans and inventory,
which means that less management is required. The reduction in levels and
numbers of managers decreases overheads, and therefore makes the companies
more competitive. As shown in Figure 6.3, planning is simplified by the reduced
throughput time because there are fewer jobs to control. Short term job sequenc-
ing can be carried out by the direct operators, while management can spend more
time on building the business and making improvements.
Accountants split operating costs into three: materials, labour and overhead.
Lean companies have less overhead, so the benefits are not so obvious. Some
managers mistakenly view an increase in changeovers as increasing direct costs,
ignoring overhead reduction in salaries, warehouse costs and obsolescence.
Line Throughput per minute Number per cycle
Item A 0.9 4
Item B 1.3 6
Item C 1.8 8
Item D 1.1 5
Item E 0.4 2
Total 5.6 25
Times per complete cycle = 4.5 minutes
Figure 6.6 Cyclic production.

96 Lean supply
6.E Developing lean operations
Plans for setting up effective lean supply should consider:
• smoothing the forecast (with the customer);

• setting up a regular demand at the forecast demand rate;
• customers’ understanding and consideration of the supplier’s constraints;
• a long-term agreement (formal or informal);
• focus on a high level of reliability (and agree what flexibility is allowable);
• mutual support and confidence;
• collaborative partnership and developments.
‘Stockless’ operations result from making other things reliable.
Although these concepts relate to external supply and demand, they apply
equally within a business, where they should be much easier to achieve. The first
key step is reliable communication. If suppliers know the volume of demand, they
can react much more quickly to individual requirements. For instance, a supplier
of painted items can paint and deliver within a few hours, as long as the demand
is expected and the choice of paint colours has been agreed previously because
capacity and supplies will have been made available.
For a fast pull-type relationship to work, the supplier has to be local to the
customer, otherwise the reactivity is not good enough. Location is of significant
importance, and some suppliers have moved to the vicinity of a major customer to
ensure close co-operation.
For lean operations, the focus should be on minimising delivery quantities, and the
means of achieving this is by reducing lead times (deliver single items when required).
Where demand lead time is always going to be greater than supply lead time, then it
is the delivery quantities that are still of paramount importance and in fact lead time
is not important (which is a common misunderstanding; see Chapter 8).
Where there is a possibility of using kanban systems, the best place to start is
with A class items. These are most important to the business, have the greatest
cost benefit and are normally among the simplest to change (while giving the
perpetrator most kudos). Close liaison is essential and involves regular friendly
meetings between people at operational level in the businesses.
Key points
• Lean principles should be adopted in all businesses.
• Lean operations are the result of improving the quality of everything.
Lean supply 97
• Lean just-in-time supply chains differ fundamentally from the

traditional ones. The key concepts are:
o pull – only buy when needed (kanban);

o one – only make or buy small quantities;
o flow – balance movement throughout the supply chain;
o quality – make processes so that they cannot go wrong;
o reliability – exceptionally good performance and communication.
• Consider the real costs instead of accounts.

• Always try to make improvements.
Note
1 MRP (material requirements planning) is used for dependent demand; see Chapter 14.
7 Safety stocks
• Using historical data.

• Demand distribution.
• Safety stock calculations – MAD and SD.
• Organising customer service.
• Determining obsolete and excess stocks.
7.A Learning from history

In Chapter 2 the balance between the conflicting requirements of good service,
low inventory costs and small operating costs was discussed. Inventory is needed
to provide this service if lean supply is not practical. Where items are required
faster than the supply lead time, the likely demand quantity is unknown and safety
stock is needed.
Three external factors shape the amount of safety stock held by an organisation:
1 The variability of demand.

2 The reliability of supply.
3 The dependability of transport.
The general approach to this situation is to set stock levels to cover the normal
variability of demand and to ensure that the other two variables are relatively
insignificant. Suppliers’ and distributors’ performance is controllable. Reliability,
not short lead time, is the key. Organise activities so that the major uncertainty is
caused by customers and their unpredictable requirements.
Stores records show the movement of stock in and out of the warehouse and
such historical information is essential for evaluating what level of stock to hold
(unless the customer provides firm orders). The quantities demanded, rather than
the number of orders, give the best guide for stockholding, but where demand
consists of a few large orders, this can be taken into consideration by either work-
ing with the customer to smooth the demand or providing supply-to-order on a
longer lead time.
Safety stocks 99
Week 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Demand 23 36 27 19 34 25 41 15 39 33 27 48 31 39 28
Figure 7.1 Demand history of item T15.
A company sells handbags to a major retail chain that orders every six
months. The order quantity far exceeds all other demands for this product,
and risking stockholding in anticipation of a next order would be a serious
burden. Therefore, instead of ex-stock supply, the retailer now waits the two
months’ lead time for delivery. If the retailer would agree to take frequent
deliveries, demand would be consistent and the supplier would be able to
provide ex-stock again, and the retailer’s inventory would be reduced.
Individual demands in the historical stock records can be analysed into weekly
totals. In fast-moving industries (e.g. fresh food) this can be daily, and where
usage rates are slow or sporadic, monthly totals are used. This process produces
historical usage statistics for each stock item (Figure 7.1).
[Note: when using demand history in forecasting, it is conventional to neglect
the demand information from the current, incomplete period unless the demand is
extremely large.]
Analyse sales into demand each week.
This data is not used directly to estimate the stock levels and supply require-
ments. More interpretation and understanding enables accurate stock levels to be
established.
The demand pattern can be displayed on a frequency distribution graph, or
histogram, which shows how many periods the demand was at a certain rate.
Figure 7.2 shows that demand was between 26 and 30 three times in Figure 7.1
and it was between 40 and 45 once. This histogram gives a good idea of the usual
demand rate and the spread of rates. In Figure 7.2 rate intervals of 5 per period
were chosen (10–15, 16–20, 21–25, etc.) and the figures on the horizontal axis
represent the maximum figures in the range. As can be seen, the most common
demand rate is 31–35.
If demand comes from a similar type of customer, then the variations in
orders will generally follow the same characteristics. These demands, although
100 Safety stocks
3
Frequency
0
10 15 20 25 30 35 40 45 50 55
Figure 7.2 Demand distribution.
0.4
Relative frequency
0.3
of demand
0.2
0.1
0
–4 –3 –2 –1 0 1 2 3 4
Variation from average
Figure 7.3 Normal distribution curve.
unpredictable, form part of a population and the characteristics of the demand

pattern are predictable. The most common demand pattern is the Gaussian or
‘normal’ distribution shown in Figure 7.3. This is an idealised form of Figure 7.2
for a very large number of periods of demand. Using the small amount of
information given in a table like Figure 7.1, the shape of the curve in Figure 7.3
can be calculated quite accurately. Demand patterns can be assumed to be ‘normal’
(even if they are not, assuming that they are normal is usually satisfactory in
practice).
The vertical scale of the graphs is frequency – how often that value occurs.
The horizontal scale is really measured from the middle (the mean) and shows the
values at which the populations occur. In practice, careful choice of the hori-
zontal grouping is needed to produce a distribution like the one shown. In general,
there is a limited amount of relevant history available, so the answer is open to
some error (where there are more than 12 time periods of data, the result is
normally good).
Safety stocks 101
As can be seen from the graphs, most of the time the demand is near the average.
Extremes only happen occasionally. A key property of the normal distribution is
that it can be described using only two parameters, the average and the average
width. Calculation of the average value is relatively simple. Measuring the width
is more difficult.
It is no use taking the highest and lowest figures, as these are usually the result
of atypical situations (stock-outs, one-off demands, holidays, etc.). It is better to
take values which are, say, a quarter of the way in from the extremes of the sample
(quartiles) or apply a more rigorous approach which uses all the available data.
Classically, average spread is measured using standard deviation (SD), although
stock controllers use a simpler measure: namely, mean absolute deviation (MAD;
as explained in section 7.B.3).
Of course, more complicated demand patterns do occur. Low demand items can
have a Poisson demand pattern, or different markets require different batch sizes
of the same product and two overlapping normal distributions exist (for example,
selling directly via the internet in single products but to retailers in pallet loads).
Assuming that the demand is normal is good enough to give sensible stock levels
in most circumstances, because the quality of forecasts then becomes a bigger
constraint.
7.A.1 Why hold safety stock?

Safety stock is the buffer between supply and demand. It decouples customer ser-
vice from supply so that each can operate independently. Safety stock calculations
cover random variations in demand. However, stock could be required to avoid
other situations, such as:
• supply failure;
• production shortfall;
• transport failure;
• slow, unreliable or inaccurate information;
• other source of disruption of service (e.g. inspection or customs delays).
As discussed on lean supply (Chapter 6), these sources of unreliability should be

managed and made insignificant. Safety stock, as calculated, will cover for these
problems, but at the expense of customer service.
7.B Normal demand patterns
7.B.1. Stock availability

The amount of safety stock will depend on the risk of running out, and therefore
the variability of demand and supply. As the actual data in Figure 7.2 forms
part of the distribution in Figure 7.3, the likelihood of excessive demand can be
calculated. This is a basic task for inventory controllers and also helps to keep
102 Safety stocks
For each item

Customer
service
Figure 7.4 Availability of a stock item.
Week 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Demand 3 10 0 135 7 19 3 70 13 0 17 54 120 2 12
Figure 7.5 Demand history of item T26.
them informed of any situations where exceptional demand will happen. Statisti-
cally, there are bound to be stock-outs eventually.
The more safety stock the better the customer service (Figure 7.4). If a stock of
200 were maintained for the item in Figure 7.1, it would probably be far too much.
The average demand was 31, but obviously some extra stock is essential; perhaps
if a stock of 50 were always available, there would be no shortages. The item in
Figure 7.5 also has an average demand of 31, but a stock of 50 would not be
acceptable. Maybe 140 would be appropriate.
This illustrates a key principle: Safety stock depends on demand variability
and not on demand rate.
People who do not understand inventory add weeks’ cover as safety stock, and
they give poorer customer service and hold inflated stocks.
Safety stock covers sporadic demand, not fast usage rate.
Using the 15 periods of data in both cases, equally good customer service can
be provided. Consider a supplier who arrives at the start of each period and tops
up the stock of the item shown in Figure 7.1 to a specified level. It has a unit cost
of £1,000. If the stock at the start of each period is just the average usage of 31,
then the stock would be expected to run out during half the time periods (since the
demand exceeds the average half the time – seven out of the 15 times in the sample
data). If there were no stock at all then the stock would run out every time and, con-
versely, if there were an infinite quantity in stock there would be 100% availability.
As the stock increases, the availability increases and this relationship is shown in
Safety stocks 103
100%
90%
80%
70%
Availability
60%
50%
40%
30%
20%
10%
0%
–3 –2.4 –1.8 –1.2 –0.6 0 0.6 1.2 1.8 2.4 3

Variation from average
Figure 7.6 Stock and availability.
Figure 7.6. As safety stock is added above the average demand, the availability
increases rapidly, but once the availability becomes high, then it requires a lot of
stock investment to improve the situation.
The amount of stockholding can be determined by statistical methods which
rely on history to predict the future and assume no change in circumstances during
periods ahead.
The variability of demand can be measured in many ways, but the best ways of
calculating it are SD and MAD. They are two ways of measuring the same thing
(just like having the price in US dollars and UK pounds). There are relative merits
for both measurements. SD is used throughout statistical analysis for all applica-
tions and is well known. It is theoretically the better measurement, but it is more
difficult to understand and work, even with software help. MADs are much more
user friendly. They are easy to understand and arguably easier to bias (the reason
for wanting to bias them will be discussed in Chapter 12). MAD is used exten-
sively in inventory management and is often preferred in practice as a successful
measure. These measures are the basis for calculating safety stocks correctly.
Use SDs if they are in the software, or if you like maths.
7.B.2 Standard deviation

The basic premise for safety stock is that the differences between the forecast and
the actual demand are random. These random variations in demand usually follow
a ‘normal’ distribution (see Figure 7.3). This is symmetrical about the average value.
104 Safety stocks
The size of the variations differs from one stock item to another (as T15 and T26
in Figures 7.1 and 7.5). Consequently, the width of the normal distribution also
differs. This width, measured by the SD, is determined by calculating the differ-
ences between each period’s demand and the average (call this the error, x), then
summing of the squares of x. This total is then divided by the number of intervals
(one less than the number of periods, N). The SD is the square root of this:
SD = √[(1/(N–1) × Σxi2]
where xi is the error for each of the N periods. Using this calculation for T15 (in
Figure 7.1) gives 8.85 and for T26 (Figure 7.5) gives 44.05.
Where there is only a little history, then N is small and the value of the SD, or
indeed the MAD, will be relatively inaccurate. This inaccuracy (the ‘standard
error of the SD’) is given by
Standard Error of SD = √ SD.
7.B.3 Mean absolute deviation

The MAD is a simpler way of measuring demand variability than SD. The error is
measured in the same way as with SD, but now the average size of the size errors
is calculated, which is the MAD.
Use MADs for ease, or to weight data for more accuracy.
A simple example will explain the calculation:
Example
The demand for item G11 during successive periods is: 5, 14, 6, 15. The
average demand is therefore 10.
The swings about that average are −5, +4, −4, +5.
Adding up these values gives 0, which shows that 10 is the true aver-
age. For measuring the variability, it is the size of the errors, not the
signs, which matter, so forget the signs and add up the numbers. This
gives 18 in total, or an average of 4.5 for the four periods, and is called
adding the ‘absolute deviations’ or ‘moduli’.
The demand pattern for a second item, H22, was: 9, 12, 8, 11 for the four
periods of history, with the same average demand as G11 of 10. The abso-
lute deviations for H22 are 1, 2, 2, 1 and the mean absolute deviation is 1.5
(6/4). Obviously, the safety stock required for G22 will be considerably
smaller than for G11. It would be a third (1.5/4.5) for the same availability.
Safety stocks 105
This can be expressed as:
Sum off absolute deviations from mean

MAD =
number of periods iincluded in the sum ( n )
The MAD (or SD) should be built into the operating system and recalculated at
each period end when new data is available. Instead of using the formula above, it
is better and simpler to use a weighted MAD which focuses on more recent his-
tory (the exponential smoothing technique is discussed in section 12.C). This can
be done very effectively with MAD, whereas the SD has to be converted to the
variance (= SD2) in order to use the weighting technique. Exponentially-weighted
MADs are therefore preferable to weighted SDs. Their use results in improved
safety stocks and, in practice, offers a better selection of forecasting techniques to
meet future demands (see section 12.D).
It should be noted that a short cut way of evaluating SD is to calculate MAD
and multiply by 1.25 (i.e. for a normal distribution MAD = 0.8 × SD).
For seasonal demands, the variations in size of the demands are likely to be
greater at the maximum and smaller at the minimum (this is expected to be a
square root relationship). The MAD can be considered as a percentage of the
demand level. Using this ‘percentage MAD’ approach enables an appropriate and
up-to-date variability value to be maintained and provides the appropriate safety
stock levels. Exponentially weighted variances should be used, but are less com-
mon in practice. Figure 7.7 shows a simple summary of the options. There is an
arbitrary quality scale included to illustrate the relative merits of the techniques.
The historical average deviation quantities calculated can now be applied to
safety stock calculations. MAD and SD are available as pre-programmed functions
Historical evaluation technique Assessment Relative quality
Stock cover Only for the amateur 1

Gives poor service and
excess stock
MAD Simple to understand and use 10
Good in most situations
SD Universal technique 11
Should give good results
Smoothed MAD Better results 25
Beneficial when style of
demand changes
Easy to setup and calculate
Smoothed variances Good technique 25
May overstate influence of
outliers
Figure 7.7 Summary of deviation measurement techniques.

106 Safety stocks
in spreadsheets (Avedev and STDEV in Excel) but not the exponentially
weighted ones.
7.C Evaluating safety stocks
7.C.1 Risk measurement

Using MAD or SD is the only way to balance risk across all stock lines and give
predictable and equal service. It is fortunate (and a mathematical truth) that the
shape of the normal distribution curve (Figure 7.3) measured in terms of MADs
(or SDs) is always the same. The middle is the average, and as half the area is
either side of the average, then holding average stock means that 50% of the time
demand will exceed this and not be satisfied. Adding MAD’s worth of safety stock
will move along the curve to the right and improve availability (Figure 7.6). There
is, therefore, a quantitative relationship between the amount of safety stock (meas-
ured in MADs or SDs) and the availability provided. This is shown in Figure 7.8.
Desired service level Multiply SD by Multiply MAD by

(% periods without stock-out)
50.00 0.00 0.00
75.00 0.67 0.84
79.00 0.80 1.00
80.00 0.84 1.05
84.13 1.00 1.25
85.00 1.04 1.30
89.44 1.25 1.56
90.00 1.28 1.60
93.32 1.50 1.88
94.00 1.56 1.95
94.52 1.60 2.00
95.00 1.65 2.06
96.00 1.75 2.19
97.00 1.88 2.35
97.72 2.00 2.50
98.00 2.05 2.56
99.00 2.33 2.91
99.18 2.40 3.00
99.50 2.57 3.20
99.70 2.75 3.44
99.86 3.00 3.75
99.90 3.09 3.85
99.93 3.20 4.00
99.99 4.00 5.00
Figure 7.8 Customer service factors (assuming a normal distribution).

Safety stocks 107
Adding one MAD’s worth of safety stock to the average requirement increases
customer service from 50% to 79%; or when stocks are increased by one SD, the
service increases to 84%, so there is a 16% chance of stock-out. By increasing the
buffer stock to twice the MAD, service is increased to 94.5%.
Additional stockholding improves the service toward 100%, but the inventory
cost is increasingly high and extra stockholding will only be used occasionally.
Three MADs’ worth gives 99.18% and four MADs 99.93%, but these are increases
of only 4.68% and 0.75%, respectively, for the same stock value that gave 29% for
the first MAD and 15.5% for the second. Customers want 100%, but that is not
affordable. Anyway, does it matter if supply fails once every 100 years (99.98%
weekly)? Will customers notice the difference?
As customer service is the most important factor for a flourishing business, the
choice of the right customer service factor (CSF) is critical – too low means run-
ning out of customers, too high means running out of money! Businesses should
assess what they are currently achieving, what the market really requires and set a
short-term target marginally better than the current performance. When this is
achieved, the target should be increased.
An illustration using stock records

Taking an example, Figure 7.9 shows a conventional stores movement
record for item T26, with receipts, despatches and supply orders for
stores. The average usage per period is 31. If the policy is to purchase
sufficient stock to cover three periods, then the order quantity will be
93. If, by coincidence, it also takes three weeks’ lead time for stock to
be replenished, extra stock is required for periods where stock exceeds
the average demand of 31. The last column shows when and how much
safety stock is required to cover the demand pattern. Without the safety
stock, there would be stock-outs during three of the 15 weeks. The
safety stock in this case was set at 97, although 68 would be sufficient
during the 15 weeks. However, the statistics suspect that eventually
more will be required. The trigger level for ordering is 187.2 (as calcu-
lated in Chapter 8). When the stock plus the outstanding supply orders
are less than this, the stock controller decides whether an order should
be placed. This is called the ‘review level’.∗
All too frequently, an order is raised at some arbitrary ‘minimum stock
level’ that bears no relationship to the current demand profile. The MAD
calculation enables an individual safety stock to be set for each item
and used in the review level.
∗Historically referred to as ‘re-order level’.
108 Safety stocks
Week Demand Variation Delivery Stock Safety stock

number (error) quantity cover needed
200 0
1 3 −28 197 0
2 10 −21 Order 187 0
3 0 −31 187 0
4 135 104 Order 93 52 42
5 7 −24 138 0
6 19 −12 93 119 0
7 3 −28 209 0
8 70 39 Order 139 0
9 13 −18 126 0
10 0 −31 93 126 0
11 17 −14 202 0
12 54 23 Order 148 0
13 120 89 28 66
14 2 −29 93 26 68
15 12 −19 107 0
Av 31 34
Figure 7.9 Stock record of item T26.
The availability level is measured here by how often the designed stock level will
meet all the demands during the period. Other measures would be ‘how many items
were short during a year?’ or ‘how many customer orders were not completed?’.
7.C.2 Effect of supply lead time

The discussion of safety stocks so far has considered the size of each period’s
variation from the forecast (MAD) and the effect this has on availability (CSFs in
Figure 7.8). The third major factor is how long supply takes (supply lead time).
It is obvious that the longer this lead time is, the more risk and therefore the more
safety stock is required. It is equally obvious that if you do not buy food on
Monday, you will have to get it some other day. Short-term demand may be erratic,
but it gets smoothed in the longer term. This means that although safety stock
increases with lead time, it does not increase at the same rate. Because of the
statistical compensation (daily demand is very sporadic, but monthly it is
smoother) the safety stock only increases as the square root of the lead time. The
relative sizes of safety stock compared to a week are illustrated in Figure 7.10.
Taking account of the three factors, the formula is
SS = CSF × MAD × square root (lead time)
(CSF is chosen in Figure 7.8). This is a practical working formula, proven many
times to give good stock balance.
Safety stocks 109
Supply lead time Relative supply lead time Relative safety stock
1 day 0.2 0.45
1 week 1.0 1.00
2 weeks 2.0 1.41
1 month 4.2 2.05
3 months 13.0 3.61
6 months 26.0 5.10
1 year 52.0 7.21
Figure 7.10 Lead time multiplier for safety stock.
The formula is the professional way to improve availability.
7.C.3 Additional safety stock

The above calculation is based on unknown demand. The quality of supply from
manufacturers may also prompt an increase in safety stock, because the
• lead time for an item varies between placing one order and the next;
• supplier’s delivery performance against the lead time is unreliable and
delivery is more often late than early;
• supply may be rejected because of poor quality (which is worse for bulk
supply when it is a long time before the next batch arrives).
In the short-term, an allowance in stock has to be made, or suppliers should

hold consignment stock. In the long-term it is unacceptable and sanctions should
be imposed.
Suppliers, not customers, should pay for poor performance.
Where stock has to be held against poor supply, theoretically a supply perform-
ance MAD could be calculated and a safety stock worked out. It would not be
additional to the demand safety stock, since the same safety stock would cater for
either high demand or late delivery, but not both. The additional safety stock could
be calculated using Pythagoras’ Theorem on the MADs (similar to adding
variances for SDs).
The suppliers’ delivery pattern should be a compromise for both organisations.
Whoever causes the need for extra safety stock should incur the cost of it.
110 Safety stocks
Minimum stock is held when each party can trust and accept the information
provided by the order. Poor co-ordination leads to extra costs, not only because
extra money is invested in inventory, but also because higher stock increases the
risk of the item being superseded or becoming obsolete.
These calculations are the basis for professional management of inventory and
are the foundations of inventory managers’ expertise.
7.C.4 Slow-moving items

For a large number of stock items, the usage rate is low. This includes obsolescent
items, non-standard lines and spares. For these, the ‘normal’ distribution is
squashed against zero usage and is replaced by a skewed distribution, typically a
‘Poisson’s distribution’. Compared with normal distribution, the Poisson distribu-
tion has a longer tail, giving an enhanced (but small) chance of high demand
during a period. To cover for this, a higher stock would be required where high
levels of availability are needed.
Demand for slow movers is a Poisson distribution.
An important feature of the Poisson distribution is that
Standard deviation = square root of average demand.
This relationship simplifies the safety stock formula for slow-moving items. Actu-
ally, this relationship is based on ‘occurrences’ not quantities (i.e. an order for two
counts as one ‘occurrence’). Therefore, the true formula is
SD = square root of average ‘occurrences’
7.C.5 Excess and obsolete items

In practice, stocks are never balanced ideally and some items become excessively
slow moving. Ultimately they are either ‘excess’ or ‘obsolete’. This obsolescent
inventory is an area which generates little interest and lots of cost. The definition
of obsolete should be that there is ‘no foreseeable use for the item’. The definition
of excess is that the ‘stock cover is greater than x weeks’. The choice of ‘x’
depends on the item type and the company policy.
Goods are classed as obsolete as a result of a long period with no demand. So
the practical definition is usually: an item is obsolete if it has not moved for ‘x’
months (Figure 7.11). Only the items in the square before x and y are under
control.
Safety stocks 111
Excess
Number of
weeks (cover) Working
Obsolete
of inventory inventory
Number of weeks since last movement
Figure 7.11 Definition of excess and obsolete stock.
The selection of ‘x’ and ‘y’ vary with the type of item and market, and the
position in the supply chain. It allows accountants to identify how much money to
put aside to cover likely stock disposal. Writing down inventory value is, of course,
an admission by the stock controllers that they failed. Obsolete stock occurs for:
• Standard products in excess caused by a combination of inaccurate fore-

casting, unreasonable manufacturing scheduling, unreliable suppliers and,
especially, bulk purchase. The solutions here are minimising supply and
promoting demand.
• Special products for single or few customers which are superseded. The solu-
tion in this case is to come to a financial agreement prior to supply, or ensure
inventory can be sold to the aftermarket.
• Manufacturing raw materials and components caused by excessive procurement
or production quantities. The solution is to reach agreement with existing
suppliers or alternatively forge agreements with new partners.
• Any product through poor control, training and focus.
Manage inventory so that excess stock does not happen.
The method for dealing with excess and obsolete stock is discussed in
Chapter 4. It is relatively easy to get rid of fast-moving consumer goods which
have not moved for three months. Alternatively, suppliers of heavy machinery and
spare parts, where world demand has been zero for two years, may not discard
items and will still desire to keep stock (perhaps only one). In this case, failure to
supply quickly may have important strategic or economic effects, especially when
supply lead times are long. A part for a machine that fails every seven years or so
is not obsolete, but has a very low usage rate. The decision to purchase such slow-
moving stock is one problem, the decision to throw away existing stock is entirely
112 Safety stocks
another. Each company has to decide on a proper obsolescence policy for dispos-
ing of non-moving items.
The first step in dealing with the phase-out of stock lines is to identify early the
lines in this terminal condition and include them in a special ABC class such as
‘O’ or other suitable flag to inhibit re-ordering.
For inventory control, obsolete stock has its cost value: accountants depreciate
the value. Any disposal above this book value will reinstate some of the profits
already lost. Inventory must be valued financially by its ‘realisable value’.
Overstating the value is illegal. Obsolete stock is the least valuable inventory.
The level of obsolete stock can be considered as a monitor of the effectiveness of
the company’s control systems. Low obsolete stock indicates better company
control systems, but more obsolete stock indicates that a company is wasting
money.
Key points
• Safety inventory compensates for unknown demand.
• Supply problems should be removed, not buffered.
• Demand can be split into:
o average;
o variability.
• Variability (safety stock) should be calculated using MAD (or SD),

but not estimated.
• Customer service factors should be used to decide the level of
availability.
• Obsolete and excess items should be identified and managed.
8 Setting the right stock levels
• Successful calculation of review level.

• When to expedite.
• The effect of supply lead times.
• Supplier delivery patterns with safety stock.
• Co-ordinating supply with target stock levels.
8.A Simple assessment of review levels

Our aim in inventory control is to provide a consistent service to customers by
maintaining each stock item at an appropriate level. The most effective time to
influence this is when committing more demand on suppliers (the reason for
excess stock is because someone has ordered it; the reason for shortages is that
someone did not order enough – beware). Ordering is when the major opportunity
occurs for ensuring a balanced inventory; it is the major control mechanism.
There are two aspects to re-ordering to consider, namely, when and how much.
This chapter considers the ‘when’ and Chapter 10 considers ‘how much’. The
technique for calculating the right time to order inventory is crucial to balancing
service and inventory investment.
In the nineteenth century, stores systems used to include ‘minimum’ and
‘maximum’ stock levels for control. This was then replaced by the concepts of
re-order levels and order quantities. This was probably prompted by confusion
over the term ‘minimum’ since the stock is at minimum immediately before delivery,
but what is required is a re-order level to identify that an order needs to be placed.
This century, supply chain operations have changed the way inventory is organised
and the focus is on ‘review levels’ and ‘delivery quantities’. These are the basic
tools for controlling individual stock lines (see Figure 8.1).
The review level is triggered when analysis of historical demand shows that
stock is low (i.e. it has fallen below the ‘re-order level’) and the controller then
decides whether to order. The ‘review level’ concept is introduced because this is
a key decision. The controller has to decide whether the future demand will be
different from the history, so the supply should be increased or delayed. If the
Order
Order
Delivery
Delivery
Supply
lead time
Quantity Review level
in stock Delivery
quantity
Use
Safety stock
Time
Figure 8.1 Inventory level.
Item T15 Item T26 Item T37

Demand history Demand history Demand history
140 140 140
90 Average 90 90
40 40 40
–10 –10 –10

1 3 5 7 9 111315 1 3 5 7 9 111315 1 3 5 7 9 111315
Item Average demand MAD

T15 31 7.1
T26 31 34.0
T37 31 20.0
Figure 8.2 Items with average demand of 31.
demand level or supply lead time is changing, the first action is to change the
review level (RL). After that, should supply still be required at this time, new orders
are placed. If demand decreases, no supply is necessary; only where the RL is the
same or higher should more supply be triggered.
If the demand for an item is 31 per week and the supply lead time is three
weeks, there had better be 93 either in stock or on order or else there will be a
shortage. As demand is sometimes greater than 31 (Figure 8.2), some safety stock
is advisable. Written as an equation this is
Setting the right stock levels 115
review level = demand rate × supply lead time + safety stock
or
RL = forecast × LT + SS
The safety stock calculation discussed in Chapter 7 showed that the stock depends
on the variability of demand (mean absolute deviation (MAD) or standard devia-
tion (SD)), the availability level required (customer service factor (CSF)), and the
supply lead time. The formula then becomes
RL = forecast × LT + CSF × MAD × square root (lead time)
Three different demand patterns are illustrated in Figure 8.2. The demand histo-
ries for items T15 and T26 were discussed in Chapter 7. Some practical informa-
tion is now required, namely, the supplier’s lead time and the availability level to
be offered to customers.
Take a situation where the supplier’s lead time is three weeks and the availabil-
ity is only set at 90%. The customer service factor for 90% is 1.6 MADs from
Figure 7.8. Using the RL formula above gives the safety stocks and review levels
shown in Figure 8.3.
This means that if the stock of T15 falls below 113 and there is none on order,
then a new order should be placed. The condition for ordering is if
Current Stock + Supply Orders Outstanding < Review Level.
To be exact, the supply orders which count here are only those which are due for
delivery within the supply lead time. The purpose of the review level is to trigger
replenishment orders at the right time. Of course, placing orders at the right time
does not necessarily ensure that there is stock in the stores.
Re-ordering triggers new orders, but does not eliminate stock-outs.
For most businesses, 90% service is not good enough and 98% is more appro-
priate. The CSF then becomes 2.56. If the review level for T26 is 187.2, and the
Item Safety stock Review level

T15 19.7 112.7
T26 94.2 187.2
T73 56.3 149.3
Figure 8.3 Review levels.

For LT three weeks and availability 90%.
Item Supply lead Availability % Safety stock Review level

time (weeks)
T15 2 90 16.07 78.07

T15 2 95 20.68 82.68
T15 2 99 29.22 91.22
T15 4 90 22.72 146.72
T15 4 95 29.25 153.25
T15 4 99 41.32 165.32
T26 2 90 76.93 138.93
T26 2 95 99.05 161.05
T26 2 99 139.92 201.92
T26 4 90 108.80 232.80
T26 4 95 140.08 264.08
T26 4 99 197.88 321.88
T37 2 90 45.25 107.25
T37 2 95 58.27 120.27
T37 2 99 82.31 144.31
T37 4 90 64.00 188.00
T37 4 95 82.40 206.40
T37 4 99 116.40 240.40
Figure 8.4 Varying the conditions.

Note: higher review levels do not necessarily increase the stock levels.
current stock falls to 139 in week 8 (see Figure 7.9), then more supply should be
triggered. However, if an order were already placed, it would not be necessary
(139 + 93 > RL). This is the case in week 3 of Figure 7.9. The order placed in week
2 increases the total provision to 187 + 93 (= 280). However, in week 4 the total
provision is only 145, so another order is triggered.
This example illustrates the advantage of leaving decimals in the calculations,
since 187 is less than the 187.2 review level; if the RL had been rounded down, the
decision would not be so obvious (where demand is very low, it is usual to use two
decimal places).
In Figure 8.4 the effects of increasing the availability or changing the lead time
are illustrated. The increased safety stock obviously decreases the stock turn.
Increased lead time increases the safety stock (by the square root), so lead time is
not so important. The order quantity stock, which is more important, should not
be affected (as discussed in Chapter 10) although the review level is increased.
This review level calculation is the basis of professional inventory management.
8.A.1 When to expedite supplies

As demand is uncertain, sometimes there will be stock-outs, and the likelihood has
been set by choosing the availability level. 99% availability means that there will
be a shortage in 1% of the periods. Of course, if the demand in the period is 1,000
and the shortage is only five, then most of the customers will be satisfied. The orders
for those five will be let down, so few customers would probably be affected.
However, it is possible to enhance the availability quite simply by identifying when
the stock is too low and then reacting. As the stock should be replenished when the
inventory gets down to the safety stock level, the basic condition is
Expedite if stock < safety stock level
If the expediting process can speed up delivery, availability will be enhanced; if

not, with the information, alternatives can be considered. Using safety stock level
in this way is a result of personal practical experience.
As the variation in demand is largely a statistical effect, the square root rela-
tionship occurs very frequently in stock control. The above formulae control the
supply process and the resulting average stock is half way between that just before
and just after a delivery; the formula is:
Average stock =1/2 Delivery quantity + Safety Stock
or
Av Stock = 1/2 Delivery interval × Demand Rate + CSF × MAD × SqRt (Lead Time)
This shows that the major factors governing the stockholding, besides the
availability for customers, are:
• supply frequency;
• variability of demand (MAD).
Lead time is not so important because of the square root. Companies should try
to make customers take products more frequently in smaller batches, and also get
suppliers to deliver more frequently even when the lead time is long.
8.B Managing lead times

Lead time is not important once the supply has been started. In project manage-
ment, by contrast, lead time is important because supply is not repetitive – it is
always for different applications.
Lead time is the time between a shortage occurring and items being available to
maintain supply to customers. It is a balance between what the supplier wants and
what the customer wants. Powerful suppliers often specify long lead times (up to
a year) but where customers have the upper hand, they may expect immediate
supply. Both of these extreme cases increase the total supply chain costs. The
compromise should give suppliers sufficient time to supply comfortably, without
forcing the customers to commit themselves further than necessary. It would be
useful to forecast likely demand further ahead for the suppliers, but the accuracy
of forecasts does reduce rapidly the further ahead they look.
Lead times for purchasing result from production and transport times, whereas
lead times for manufacturing result from capacity, material availability and
production planning. Lead times required by customers depend upon availability
and supply policy (make-to-order, assemble-to-order, ex-stock).
8.B.1. Components of lead time

Lead time can be analysed into components, some of which are essential and some
avoidable. The components are:
• Order review time: the frequency with which the stock situation is investi-
gated (e.g. orders placed daily, hourly, weekly, etc.).
• Order processing time (purchasing and communication): the time it takes to
review inventory history;
decide to buy;
transmit to the ordering system;
raise an order;
gain the appropriate authority;
inform the supplier (media, email, physical order).
• Supplier lead time (vendors’ manufacturing, buying and despatch).
• Transport time: transfer of the item from supplier to receiving bay (including
customs and freight forwarding).
• Receiving time: time taken for goods inwards and updating the stores records.
From this analysis, the actual supply lead time contains some process elements
that add to the lead time quoted by the supplier.
Use the actual lead time, not the theoretical one.
In stock control equations, supply lead time is assumed to be constant. It is very

important that the supplier aims for consistency of lead time and the actual lead
time is used in the system. When supply lead times change, usually for groups of
similar items, then the lead times in the calculations must be changed to reflect
this. It is a continuous task and part of the maintenance of the stock systems, by
using the actual lead time to update (exponentially weighted) standard lead times.
8.B.2 Setting lead times

Long lead times and large batch supply produce an order pattern which is irregular
and subject to change. If the actual lead time is longer than that set on the system,
the number of stock-outs will increase since the stock level will be too low.
Conversely, where the system lead time is over-cautious (too long), then inflated
stock levels result. Setting the system lead time is a fine judgement for the
inventory controller.
Longer lead times make forecasting more difficult and increase safety stocks.
Reducing the lead time is therefore desirable, but not as important as controlling
increasing delivery frequency and forecasting better.
A major difficulty is often unreliable supply lead times. Taking the worst case
increases the stock level. One simple way to improve the situation can be to
specify lead time more precisely. Requesting ‘delivery in the month of June’ gives
suppliers plenty of leeway and July is not considered too bad. Requiring ‘delivery
on 12 June at 11.18 a.m.’ suggests an entirely different expectation, and is likely to
gain much better delivery performance (during the day required rather than the
month). This is borne out in practice.
If supplier lead time is long, more leeway is considered the norm, but a week’s
extra inventory held to avoid stock-out with 20 weeks’ lead time is as costly as an
extra week’s worth when the lead time is only a week.
In many cases, the lead time does not require inventory (supply lead time is often
job queueing time and a manufacturer may not buy the materials until well after
they receive an order). Internally within a business, lead time is the time between
receipt and despatch of an item: here, investment in inventory is necessary, so it is
beneficial to reduce throughput time. Techniques for achieving this include:
• inputting information into the system as transactions happen;

• minimising the time between processes;
• carrying out all processes at the demand rate;
• partnership with major suppliers (giving schedule flexibility).
Where supply can be shortened to less than the demand rate, it becomes
dependent demand and, with very short lead times, kanban pull supply can be
introduced.
Example
Our supplier was quoting six weeks to deliver to us. What did this
mean? It could be:
• The supplier always makes-to-order for this item. It is non-stock

and it takes six weeks to get these from their supplier.
• All the existing supply is allocated to someone else. There is already
a queue of orders to fulfil.
• There is a stock-out because the safety stock was not high enough.
This could be the rare occasion where there is a demand spike, or
just too low safety stock.
• The re-ordering procedure broke down. It was supposed to be

ex-stock, but the replenishment was not requested.
• Our demand is excessively high, so they hold insufficiently large
stock.
Collaboration with suppliers can reduce the effective lead time. If the
supplier is forewarned of our (regular) demand, they can get procedures
started or hold stock in advance. By discussion with the supplier, we
agreed to send a plan for regular demands which we received each week.
Furthermore, the time actually spent being processed was much
shorter than the manufacturing lead time (less than 10%). By elimin-
ating waiting time, the manufacturer reduced the throughput time and
that allowed us to alter the quantities two weeks ahead. It did take some
time to make these changes work.
Because of the large numbers of lines of inventory, lead time reduction has to
be carried out selectively. These are initially:
• A class items;
• items where a small change in lead time can make them into supply-to-order
items.
8.B.3 Influence of lead time

Lead time does not govern stockholding. Order quantity is not affected by lead
time. It is possible to order 20 weeks’ worth of an item with one week’s lead time
and one week’s worth of an item with a 20-week lead time (as long as further
orders are placed each week). True, the review level is calculated using the lead
time, but the review level determines only what order cover is required and not
how much stock is held.
For repetitive demand, reliable lead time is essential, short lead time is not.
Lead time does affect the amount of safety stock to hold. The longer the lead
time, the wider the absolute variation in demand and therefore the higher the
safety stock for a given service level. Fortunately, the amount of extra stock only
increases as the square root of the lead time. In summary, the effects of long lead
times are generally:
• forecasting demand to cover the supply period is more difficult;

• the reliability of the supplier may be poorer.
Errors in forecasting cause higher stocks to be held. If demand is certain over
the lead time, then stocks can be very low. A short lead time means that:
• the forecast is more likely to be correct;

• errors will be smaller.
8.B.4 Dealing with inconsistent lead times

With closer partnerships and higher expectations, inconsistent delivery is not the
problem that it used to be. However, customers are now more sensitive to supply
reliability, so strategies should be considered for any supplier where delivery is
inconsistent and changing supplier is undesirable.
It is difficult to set reasonable stock levels or to provide good customer
service if lead times are not known accurately. The policy on inconsistent lead
times should be determined by the impact of the problem, and on whether the
financial risk is high or low. For high turnover value items (A class), the
inconsistency has to be resolved through scheduling and working more closely
with suppliers to smooth out the inconsistencies. The objective should be to
smooth out the variations rather than to improve the updating, since each time
the lead time is updated, the stock parameters will be changed and the order
book modified.
For low turnover value items (C class), the situation is not so important unless
there are very many inconsistent suppliers. The strategy for avoiding the problem
for routine supplies is to use pessimistic lead time in calculations. This will cause
higher stocks, which may be necessary if the item is important. However, as the
purchase quantity is normally large, there are only a few occasions on which there
is a chance of running out, so C class items may not be worth the investment in
extra stock to provide the extra availability.
In some instances, lead times are cyclic, as a result of the supplier having a fixed
plan and not responding to customer demand.
Example
In many industries, there is a logical sequence of production which can-
not be broken: in steel rolling mills (thickest to thinnest over several
weeks); in injection moulding (light to dark colours); in mixing processes
(pure to impure); and many other cycles (food crops, etc.).
In this case, supply orders should ideally be placed when the lead time is min-
imum. Using target stock levels (described in section 8.D) and good communica-
tion, the timing of orders can be synchronised to accommodate any planned
changes. This is a significant challenge.
8.C Supplier delivery frequency effects
The safety stock calculated by the methods described earlier in this chapter are
adequate for most practical situations, and the following discussion adds
complexity which should be used once the simpler formulae are working. The
calculations can be improved because the risk of stock-out is not constant all the
time. Immediately after a supply delivery, availability is excellent. The risk of
running out is greatest when the stock is lowest, i.e. just before the supplier deliv-
ers, at the time when inventory is below the review level. During a year, therefore,
the main risk is during the lead times, which is the previous calculation. This
depends upon order frequency. The more replenishment orders for an item that are
placed per year, the higher the risk and the greater should be the safety stock. This
is illustrated in Figure 8.5. The increase in safety stock required is usually less
than the benefits of smaller delivery quantity stock, which reduces the overall
inventory cost and risks of obsolescence.
A: 99% availability
4.5
Customer service factor in MADs
4 4 weeks LT
2 weeks LT
3.5
8 weeks LT
3
2.5
2
1.5
1
0.5
0
0 5 10 15 20 25 30
Delivery interval in weeks
B: 95% availability
3.5
Customer service factor in MADs
3 4 weeks LT
2 weeks LT
2.5 8 weeks LT
2
1.5
1
0.5
0
0 10 20 30
Delivery interval in weeks
Figure 8.5 Safety stock and delivery frequency.

Taking account of supply frequency refines the safety stock levels.
If the delivery size lasts longer than the lead time (i.e. LT = 2 weeks, but
order = 6 weeks’ worth), then less safety stock is required, but if delivery
frequency is shorter than the lead time (LT = 6 weeks, delivery = every week),
then availability is less than the simple calculation. Where a new order is placed
when the previous one arrives, then the previous calculation is exactly correct.
Example
The annual demand is 10,400 (although the results are not specific to
this) and 99% availability is required, which means only 104 items short
during the year. If the stock is at risk every week, then a shortage of only
2 is allowed, but if the deliveries are every 8 weeks then a shortfall of
16 is allowed. Figure 8.6 shows the availability and the CSF, and the
corresponding number of MADs required, on the basis of Figure 8.5.
The safety stocks can then be calculated in the normal way.
Figure 8.7 identifies the correction to the original safety stock formula. The
horizontal axis shows how long the delivery quantity compares with the lead time,
and the vertical axis is the corresponding proportion of original safety stock that
would be required.
Availability 99% 4 weeks LT 2 weeks LT 8 weeks LT
Delivery Supply Allowed Availability CSF Availability CSF Availability CSF

Frequency Batch Stock-outs required required required
(weeks) Size each order
1 200 2 99.75% 3.51 99.5% 3.20 99.88% 3.82

2 400 4 99.50% 3.20 99.0% 2.91 99.75% 3.51
3 600 6 99.25% 3.04 98.5% 2.71 99.63% 3.31
4 800 8 99.00% 2.91 98.0% 2.56 99.50% 3.20
5 1000 10 98.75% 2.81 97.5% 2.46 99.38% 3.13
6 1200 12 98.50% 2.71 97.0% 2.35 99.25% 3.04
7 1400 14 98.25% 2.64 96.5% 2.27 99.13% 2.97
8 1600 16 98.00% 2.56 96.0% 2.19 99.00% 2.91
13 2600 26 96.75% 2.31 93.5% 1.91 98.38% 2.64
26 5200 52 93.50% 1.92 87.0% 1.43 96.75% 2.31
Figure 8.6 Effect of supply batch size on availability.

For supply of 200 per week.
160%
99% Availability
140%
95% Availability
Multiply safety stock by
120%
100%
80%
60%
40%
20%
0%
0.00 5.00 10.00 15.00
Ratio of delivery frequency to lead time
Figure 8.7 Safety stock correction for delivery quantity.
Example
If a suppler delivers every two weeks and the delivery quantity lasts six
weeks, this is three on the x-axis. For this ratio, the safety stock can be
reduced to 61% of the formula value for 95% availability, and 81% for
99% availability.
This correction becomes important when the delivery frequency and lead
times are very different, as can be the case for international long lead time supply
with weekly shipments, or large batches bought on short lead times (which is
typical for C class, and provides better availability than expected). Using the
basic formula with a target of 95% availability, buying ten weeks’ worth on a
one-week lead time would actually only require a quarter of the safety stock to
give 95% (Figure 8.7).
8.D Target stock levels

Businesses try to buy more of their supplies from fewer vendors, which means
that buying them all at the same time is more efficient. One way of achieving this
is to set a top target stock level (TSL), sometimes called an ‘order-up-to’ level.
With TSLs, orders for a variety of items are placed at the same time, often regu-
larly on the same day every week, fortnight or other intervals (hence ‘periodic
review system’). The workload of the inventory purchaser can therefore be organ-
ised and balanced more efficiently.
As TSLs fix the time between orders, the amount ordered varies depending on
the recent demand. With review levels, the order quantity is stable and reflects
forecast demand. The timing of demand varies, which could result in spending
time creating a succession of individual orders from the same supplier. The TSL
approach is therefore good to use for the regular demand items, and is recom-
mended for A class products.
8.D.1 Application of target stock levels

The TSL is a level to which the stock is topped up (theoretically) when the cycle
time for replenishment arrives. The process for setting up the ordering procedure
is to agree with suppliers:
• the cycle time for ordering;

• the day on which TSLs are reset and orders are reviewed;
• the day when deliveries will be received (one lead time later);
• the flexibility for order quantities.
The amount which is to be ordered is given by the formula:
order quantity = TSL – free stock – supply orders outstanding
Here the ‘orders outstanding’ are, of course, only those orders which are due
within the current supply lead time. Schedules and orders booked further ahead
are not included in the calculation. This is illustrated in Figure 8.8.
New stock is ordered at fixed points in time
Review Review Review Review

cycle cycle cycle cycle Target
stock
2,000 Pre-determined stock level level
Stock level (units)
800*
1,500
1,300*
1,000
Stock level
500 * Delivery
quantity
lead time lead time lead time
0
1 2 3 4 5 6 7 8
Time (weeks)
Figure 8.8 Periodic review level.

When buying a variety of items from a supplier, use TSLs.
If the lead time is longer than the review cycle, which is often the case, the
calculation must include the outstanding supply orders as in the formula
(see example).
Example
TSL calculated as 34
Current free stock 16
Outstanding order due tomorrow 5
Therefore
Replenishment quantity = 34 − 16 − 5
= 13
8.D.2 Calculation of target stock levels

The TSL is simply safety stock plus a standard delivery quantity. This delivery
quantity is the forecast demand over the selected number of weeks’ cover. The
safety stock now has to cope with the period until the review cycle comes around
again, as does the delivery quantity. The formula is
TSL = [Usage rate × (Order cover + Review Period)] + TSL Safety Stock
where
TSL Safety Stock = CSF × MAD × Sqrt (Lead Time + Review Period)
The demand ‘in review period’ is included because replenishment is only

triggered at the review time. Where deliveries are also cyclic (e.g. once per
fortnight), this time interval should be added to the lead time as well.
Example
Review period for an A class item = 1 week
Order size in weeks = 2 weeks
Supply lead time = 5 weeks
Average demand rate = 4 per week
MAD = 2.5 measured on a week basis
Required customer service = 90% (this gives a CSF of 1.6)
Therefore,
Safety stock = 1.6 x 2.5 x √ (5 + 1)
= 9.8
and
TSL = 4 x (2 + 1) + 9.8
= 21.8
and there will always be two orders outstanding, so the stock will be
well below the TSL.
The TSLs are levels of stock to which the free stock never rises in a stable situ-
ation (see Figure 8.8). The actual stock will rise to a maximum level, which is:
TSL − (usage rate × lead time). In fact, where the delivery quantity is less than the
supply lead time, there are more than one replenishments outstanding, so the
actual quantity in stock is much less than the TSL.
As with review levels, it is important to keep on adjusting the TSLs to match the
current usage forecasts and lead times, otherwise this approach can cause last
minute purchases and extra work. This can be done when preparing the replenish-
ment each cycle; however, it is useful to continuously monitor when the stock falls
below the safety stock level in case urgent out-of-cycle replenishment is needed.
Key points
• Review levels should be continuously updated in line with demand.
• Review levels do not avoid stock-outs; to do this requires expediting
when physical stocks get low.
• Batch sizes are more important than lead time in keeping stocks low.
• Using target stock levels is good when buying many items from one
supplier.
9 Procurement
• The role of procurement.

• Supplier relationships.
• Single sourcing.
• Vendor evaluation and rating.
• Supply partnerships.
9.A The role of supply chain procurement

The development of supply chain management changes the relationships between
companies providing goods to one another. This has affected the role of the procure-
ment function. Purchasing has traditionally been an order-by-order process focused
on obtaining the best combination of price, quality and delivery. Expansion into
strategic procurement has changed these concepts. Procurement is now involved
with all aspects of business linking to quality, design, capital expenditure, outsourc-
ing strategy and even marketing. For inventory management, procurement means
organising repetitive lean supply through contracts, which is the opposite of the old
attitude of ‘bulk buy to reduce unit cost and inventory is someone else’s problem’,
although the old ‘three quotations and the best wins’ approach is still relevant when
buying for projects.
In supply chains, the role is to create alliances with customers and suppliers for
mutual benefit. Providing end users with ideal products and avoiding waste is
more important than unit price reduction. The focus on price reduction and short
lead times has been replaced by long-term alliances, reliable performance and
small delivery quantities.
The first step is to understand the position relative to vendors. Some items are
difficult to get while others are freely available ex-stock. Some suppliers are easy
to work with, and others very difficult. In order to understand better likely causes
of this, Kraljic1 considered supplier and customer attitudes to supply of a product
using the normal four-box classification. Kraljic’s Matrix is interpreted for inven-
tory purposes in Figure 9.1. For effective supply chain, Strategic and Standard
boxes are comfortable. Strategic is best, especially for A class items.
Procurement 129
Leverage items Strategic items

Description Customer controls Description Products important to both
Important
supply parties
Relationship Small supplier reacts Relationship Collaboration, mutual
to large customer’s whims dependency
Solution Benevolence, planned Solution Single source, communication,
Importance to customer
demand kanban
Situation Supplier has to be agile Situation High volume A class items, low
and therefore inefficient stock
Standard items Risk items
Description Standard products Description Vendor dictates supply to
Unimportant
ex-stock, many potential customer

suppliers Relationship Large supplier, small
Relationship Not close unless customer Usually poor delivery
price advantage performance
Solution Rationalise vendors, Solution Seek alternative supply
contracts, VMI Situation Patented product or bulk
Situation C class items, use manufacturer
alternative suppliers
Unimportant Important
Importance to customer
Figure 9.1 Kraljic’s Matrix.
The Leverage box is often a problem for the supplier as they can be entirely
dependent on the fluctuating demands of one customer. The customer has to con-
sider the supplier’s constraints and either ensure that the supplier will be capable
of providing the appropriate volume and variety in the long term, or phase out and
find an alternative. Many major companies (e.g. supermarkets) like the Leverage
box as they have control of the supply chain, but it puts major strain on suppliers
(the original Toyota kanban model was based on their dominance).
Conversely the Risk box is where the customer needs the product, but the sup-
plier is not bothered. Customer care and delivery performance can be poor. In the
other boxes, a policy can be imposed where ‘the perpetrator pays for poor perfor-
mance’, but not here. The product is either low volume, peripheral to the supplier’s
business or a nuisance to provide (often exacting quality). The customer has there-
fore to avoid the item (design out) or is forced to hold excessive inventory. Where
users specify Risk box products, the extra costs should be borne by the user.
Since Kraljic’s original work, these ideas have been developed so that they can
be applied in different operating situations. The importance for the manager is to
understand which box they are in during a supply negotiation, so that the appro-
priate attitude can be adopted.
At some stage in the supply chain, items have to be manufactured. The manufac-
turing attitude is often critical to the inventory in the whole supply chain. Primary
processes usually produce a narrow range of commodities in large batches and
gradually along the supply chain the diversity of items created by this commodity
increases while the demand decreases until at the user end there is infrequent
130 Procurement
Manufacturing attitude Customer attitude Compromise

Provides products Requires solutions Supplier increases profit
thought solutions
Supplier wants narrow Customer sources wider Increased range secures
range of options range from each supplier customer
Bulk low cost production Require time-phase deliveries Lean production master
schedule
Own design, efficient to Fits in with products from Demand sensitive design
produce other suppliers
Fill capacity Consistent lead time Manufacturer adopts lean
supply
Delivery when available Fast reliable service, ex-stock Agreed delivery cycle plan
with lead time where possible
Long-term fixed plan Flexibility, respond to Schedule with flexibility
reasonable demand changes options
Generic service Different service for each Develop agile processes
customer
Make margin Reduce cost Improve efficiency & design
Avoid waste Reliable products Quality systems operating
Efficient manufacture Small batches Reduce production
change-over times
Minimise cost to fulfil Convenient ordering process Integrated systems, VMI
demand
‘Provide products’ Packaged items ready for Customise products
philosophy customers to use or sell
Figure 9.2 Supplier attitudes.
demand and wide variety. The differing standpoints of manufacturers and their
customers are summarised in Figure 9.2.
A significant shift in the attitude of manufacturers is happening because global
competition enables more customer choice, so change is essential to remain com-
petitive. The onus is on manufacturers to make the majority of changes, although
these are mainly conceptual and not costly. Lean high-quality supply has a major
effect on the type of expertise required in purchasing because low-level ordering
is being replaced by schedules and high-level contracts. Communications technol-
ogy enables inventory managers to schedule supplies directly from their systems,
avoiding processing delays and errors. This requires a lean environment. Ideally:
• total quality – supply to specification is expected;

• single sourcing – only one supplier for each item;
• supplier partnerships – integrate activities with supplier;
• logistic control – co-ordination of production warehousing and transfer costs;
• flow – at all stages, supply at the consumption rate;
• supply chain management – maximise effectiveness and minimise total supply
cost.
Procurement 131
The role of procurement is contract negotiation – a skill which inventory man-
agement has been practising internally for a long time. Purchasing no longer
requires the wheeler dealer, buying from a favoured source, hard hitting on price,
and always with a good idea of where to go for a cheap deal or a quick supply.
Procurement is contract management, working in partnerships and shrewd enough
to balance the wellbeing of their company with the aspirations of their supply
partners. They will have fewer suppliers and maintain multilevel collaboration
within the supplying and user companies. In some companies, top executives
negotiate the purchasing agreements, as they did in the early days of business
enterprise. This situation arises because of the improvement in communication
between companies and the better organisation of the supply chain.
9.B The purchasing environment

The environment in most businesses has changed from a supplier-led to a customer-
led operation. This has resulted from the increase in international trading, multina-
tional competition and improvements in distribution systems.
The development of the customer-led market has caused major changes in sup-
ply structures. There have been significant developments. The current climate for
sweeping inventory out of the supply chain, despite much international long lead
time supply, has been brought about by businesses adopting lean processes:
• Supplier relations are much closer.

• Quality management works on audit, not inspection.
• Stock control philosophy is to keep low inventory.
• Supply chain structure is consistent.
• Supply is consolidated, giving closer relationships with fewer vendors.
• Distribution is reliable and costed directly, not considered an overhead.
• Communication is instantaneous and reliable, enabling tighter targets.
Instead of fighting over price, the discussion is about co-ordinating and balancing
inventory along the supply chain.
Work more closely with fewer suppliers.
The process for developing a new supply partner is usually:
• select vendor;
• assess expertise;
• accredit;
• agree supply type;
• test performance;
• set up agreement;
• monitor performance.
132 Procurement
This monitoring is then fed back to the supplier and maintained on a continuous
basis.
9.C Single sourcing

Quality is king! In order to ensure reliable quality of product and delivery, pro-
curement has had to increase the co-ordination with suppliers and consequently
reduce the number of suppliers. Single sourcing and buying more variety from
each supplier can achieve this.
Where supply reliability and cost are never a problem, then there is no need for
dual sourcing. The aim is therefore to work towards this situation. To clarify the
definition of single source it is where ‘an individual product is sourced for only
one supplier’. This does not mean putting all your eggs in one basket, but it does
mean relying on a supplier for a particular size of egg. This always gives the
back-up that the size 3 egg supplier could be persuaded to provide size 2.
The approach towards single sourcing is a typical lean improvement process:
• improve the process until a problem is encountered;

• identify the cause of the problem;
• find the means of solving or avoiding the issue;
• advance to the next problem.
Managers often consider that their own situation is not appropriate for single
source supply. Traditionally, companies purchased items from a variety of suppli-
ers, choosing a supplier at any time on the grounds of cost or availability, or less
often, quality. Such formal, arm’s-length relationships, with the customer giving
‘orders’ to the most fitting suppliers, was found to be more time-consuming,
costlier than average and less reliable than commitment to one vendor.
Where there is no customer commitment, the supplier will:
• present a higher price, then negotiate;

• only meet the customer’s specification and no more;
• offer improvements only under duress;
• provide add-on services at a cost;
• need to profit on that order;
• fear poor repeat business.
This negative attitude stops the supply chain from being co-ordinated properly
(unless they are leverage items). Traditionally, managers worry about single sourcing
because of:
• too much supplier power;

• inconsistent quality;
• gradual price inflation;
• unreliable delivery performance;
Procurement 133
• lack of flexibility;
• supply ceases (through policy or liquidation);
• sensitive information security.
These are real fears for those not accustomed to single sourcing and present the
first barriers to be crossed. Suppliers may be wary of committing major capacity
to one customer.
Focus on least total operating cost, not lowest unit price.
Companies now realise that they are able to reduce these risks and can move to
single sourcing with confidence. The benefits are shown in Figure 9.3. It enables
a transformation of relationships, lowering of inventories and lean operations.
If suppliers are confident of a significant and consistent demand, they will develop
more efficient specialist processes and become more committed to customising
their product, service, flexibility and communication for their customer. Most
importantly, single sourced providers can develop new processes with less fear of
competition knowing about it.
To make single sourcing work in practice, both managements have to be con-
vinced that cost savings from these benefits far outweigh the potential risks involved.
Initially choose sources that are profitable and where the level of demand is of
significant value, and then make sure that performance is continuously excellent.
Take the supplier’s point of view.
Single sourcing also means empowering the supplier to provide the quantity,
quality and delivery without fail. Entrusting this supply and agreeing performance
benefits or penalties ensures success (and punitive penalties in the case of the
automotive industry). The key to success is for the provider to commit to supply,
not necessarily to produce, but to be responsible for a back-up supply should prob-
lems occur. This subtle change makes single sourcing widely acceptable.
9.D Supply partnerships

For effective supply, companies need support from both suppliers and customers.
The most effective way, particularly in the upper part of the Kraljic Matrix, is
through partnerships. There are many interpretations of the ‘partnership’ concept.
Although it is based on integrated operations, this can be anything from kanban,
co-ordinated scheduling, collaborative developments and joint ventures, to full
financial openness and co-operation.
134 Procurement
Supplier Customer
Higher share of business Reduction in buying and paperwork

Better opportunity to plan resources Less vendor management
Confidence in continuing demand Easier traceability for non-conformances
Opportunity for product and process Single shipping route
development
Regular demand Consolidated deliveries
Advantages
More consistent quantities Faster replenishment cycle

Improved planning One supplier responsible for each item
Better opportunity to plan resources Joint quality improvement processes
‘Base load’ business Lower stock through co-ordinated supply
Opportunity for added value services Higher volume so more buying leverage
Repetitive consistent demand Supply focused on quality and efficiency
Supply can be lean Joint cost reduction programmes
Understand customer demand better More stable schedules
Closer communication
More dependence on one customer Supplier has too much power
Need to divulge financial information Risk of poor delivery performance
Sales volume depends on the Lack of flexibility
customer’s success
Risks
Customer may change supplier Supplier may cease trading

Customer may run into difficulties Divulging sensitive information
Expertise may be leaked to competitors Risk of quality problems
Customer will squeeze price Supplier may increase price once
established as source
Figure 9.3 Single sourcing advantages and disadvantages.
The development of mutual confidence and a close working relationship enable

partners in the supply chain to plan facilities to match the ongoing demand, focus
on quality and delivery (just-in-time), eliminate formalities (time and paperwork)
and provide a service rather than simply a supply of items. The aim is to optimise
added value throughout the supply chain to the benefit of both partners. This
requires a mutual commitment to collaboration over several years towards:
• long-term collaboration;
• open relationships;
• mutual problem solving;
• supply chain management;
• quality conformance;
• collaborative development.
Procurement 135
Philosophy Collaboration Development Quality Finance and Physical

management value supply
Similar Customer Resourceful Standards Price Frequency
mission focus and and batch
and vision accreditation quantities
Legal and Understanding Flexibility Product Commitment Reliability
ethical reliability and timing
values
Financial Back-up Technical Product Cash flow Communi-
stability support lifetime cation and
information
sharing
Size and Integrated Customisation Quality Product Lead time
capacity interface management style
processes
Lean values Marketing and Growth Achieving Value Geographic
promotion excellence focused location
Trustworthi- Processes and Process and Audit and After sales Handleability
ness structure product administration service and
development packaging
Figure 9.4 Partner selection checklist.
There are also some practical considerations, including geographic location,

communications, flexibility and company philosophy. Figure 9.4 is a checklist
used in practice.
Partnerships provide long-term success.
For partnerships to work there should be a long-term match in businesses and

aspirations, the overall style, quality standards, ethics, capability, capacity and
development potential. Partnerships can improve product flow tremendously, and so
reduce inventory and increase flexibility and profits. Improved mutual trust means
shared risk, better forecasting, less safety stock and availability of stock records to
each party. Successful partnerships agree on attainable targets that stretch both part-
ners to give equal benefit. Additional benefits arise through sharing development
skills and operational and financial data. Partners should have similar views of their
roles and futures.
9.E Vendor appraisal

The traditional purchasing targets of good quality, price and delivery should be
interpreted to mean total quality, long-term total cost reduction and consistent or
just-in-time supply.
136 Procurement
Suppliers are responsible for providing good quality products that meet per-
formance, aesthetic, reliability and durability standards. Quality also means safe
delivery, on time, accompanied by accurate information on packaging (including
the customer’s identity code and shipment quantities carton by carton).
Traditionally, it was normal practice to inspect all goods arriving, or on a sam-
ple basis. As it is customary for suppliers (particularly manufacturers) to inspect
goods being despatched, items are therefore inspected on despatch from one com-
pany and inspected on arrival at the next. This infers that the supplier or transport
company is failing to meet required standards and there is a degree of mistrust.
There are always stories of suppliers sending wrong quantities or wrong items,
even though the consignment may have been inspected on despatch, particularly
if the supplier is also the manufacturer. The integrity of the method of supply may
also be questioned, more so if it involves international or long-distance transpor-
tation. If customers inspect deliveries and find problems, it will lead to complaints
to the suppliers and the questions which then arise are:
• Who paid for the inspections and when?

• Who separates the defective parts?
• Who pays for the return transport?
• Who owns the problem?
• Who pays for a vendor rating and quality control system?
Make suppliers pay for the cost of their quality.
In most cases the answers to these questions is ‘the customer’. If the supplier is
not worried about this aspect of customer service (especially for a single source or
accredited supplier) getting the customer to carry out their quality assessment seems
to be a good cost saving idea. Only where the supplier’s business is jeopardised by
the quality situation is the matter resolved. The options for the customer are:
• Send the whole shipment back (‘but we need it now!’).

• Establish and circulate a league table of supplier ratings.
• Charge suppliers for the cost of inspection.
• Communicate serious problems to their chief executive.
A business can only meet its own customer requirements if it has reliable sup-
pliers (or holds extra inventory, which is unacceptable). If supplier reliability is
missing, the business must make changes to get better service.
The reason for getting into this situation is choosing inferior vendors in the first
place. It is their responsibility to ensure that defective items do not get through to the
customers. The best situation is where there is no need to assess supplier perform-
ance because they do it effectively and cause no problems. This is why quality
management systems accredit company process capability, not just products.
Procurement 137
Audit is a necessary part of vendor assessment, to ensure that suppliers are
capable of maintaining excellence of supply. The purchaser’s appraisal consists of
three major areas:
1 Technical: is the item fit in form and function?

2 Supply: is the quality, quantity, timing, identification and cost acceptable?
3 Support: is the supplier assisting in developments, queries, re-engineering?
The detailed criteria in each area differ between businesses. They are developed
by the management team and ranked in importance (some features are essential,
while others are just advantageous). The historical, or potential, achievement
against these criteria is then assessed and a supplier rating given. Vendors can be
broadly classified into the categories shown in Figure 9.5, which then define the
inspection regime.
• Certified vendors: Businesses should attempt to have all their key suppliers in
this class, attaining reliable, self-assessed performance.
• Qualified vendors: The majority of vendors are normally in this category,
where performance is not perfect and so some inspection cost is incurred.
• Approved vendors: This includes other suppliers who have been assessed, but
reliable performance history is scarce or poor.
• Proscribed vendors: These companies may be disallowed for legal, competitive
or political reasons, or because of past lousy performance.
This vendor rating system can be based on a programme of liaison visits to

vendors to ensure that their operations are reliable and exceed the quality stan-
dards required for this supply. In this process, new quality initiatives and any
remedial actions should be agreed. These visits are routine and not responding to
shortcomings (the purpose is to avoid these). The results of these regular audits
are shared with the suppliers and actions followed up. The supplier classifications
should be applied generally with occasional exceptions where it is essential.
Vendor class Status Inspection

Certified First choice; close relationship Capability audit
Trusted suppliers; industry accreditation No inspection
Qualified Select group of suppliers Audit and sample inspection
Business quality audited
Offered expanding business
Approved Used occasionally for standard items Conformance certification or
Items not available elsewhere full inspection
Prototypes, non-inventory items, office
supplies
Proscribed Avoid using except where unavoidable 100% detailed inspection
Figure 9.5 Classification of suppliers.

138 Procurement
Attribute Importance I Achievement A Score I × A

Delivery on time 35% 90% 31.5%
Product failure rate 35% 95% 33.3%
Aesthetic appearance 10% 100% 10.0%
Convenience 5% 30% 1.5%
Information and tracking 5% 80% 4.0%
Style and problem resolution 10% 70% 7.0%
TOTAL 100% – 87.3%
Figure 9.6 Vendor appraisal.
It is useful to have a specification of these categories and their importance

for the customer and conditions under which suppliers would be abandoned or
supported.
In Figure 9.6, some key attributes for a regular supplier are illustrated, along
with the relative importance of each and the achievement of the supplier against
each.
The relative importance of the attributes is somewhat subjective and needs to
be carefully constructed. Some aspects of supplier performance against these
can be monitored quantitatively, while others are also subjective (and depend on
who is assessing them). However, a final success rate can be generated (87.3%
in Figure 9.6) and serve as a basis for improvement in co-ordination. Each busi-
ness should define its own key attributes and the relative importance of each.
The achievement marking should also be carried out by a team to avoid personal
bias.
Vendor performance measurement is primarily to create good and improving
supplier performance. Agreements on financial rewards and penalties are very
useful to make it work in practice. The supplier could receive a premium if the
rating is excellent, and the purchaser gets a discounted price if the rating is poor.
The overall effect of this can be to reduce the total quality cost between the two
companies.
Careful vendor selection results in reducing the number of suppliers and work-
ing more closely with them over a long period. With appropriate performance
criteria and incentives, suppliers can achieve certified status and thereby reduce
everyone’s costs.
9.F Pricing methods

Formerly, purchasing had a fixation on price, but now it is obvious that reliable
delivery and high quality are more important. Suppliers want to sell goods to their
target market, and purchasers will get a better deal if they are within it. This target
could be a short-term arrangement (promotional, securing market share or dis-
posal of excess), so it is usually advantageous for the purchaser to arrange a
Procurement 139
long-term agreement. Vendors want to incentivise customers to comply with their
target strategy, so purchasers who identify this can take advantage.
Prices should reflect how to make the whole supply chain prosper.
Purchase price may constitute 50% of the sales value, but a lost sale is 100%.
A contract price should be fair to both parties and encourage long-term improvements.
Ways of arranging the price include:
• market price – at time of ordering;

• inducement price – to attract customers;
• fixed term contract – competitive value choice;
• market economic price – inflation agreed formula;
• contract price with discounts (value, volume, variety).
The total cost for supply from each vendor is much more important than unit
cost of each line. Suppliers’ strategy may not be about short-term profit, but long-
term goals such as entering a new market, increasing market share or promoting a
brand. They may provide standard products at low cost, but charge extra for cus-
tomisation or added value services, and the purchaser then has to consider the
total cash flow implications in order to make the correct decision. The purchaser
has always to consider optimising the value to their business, not just cost.
Key points
• Procure by creating strategic contracts for scheduled supply.
• Identify supplier power and select vendors carefully.
• Close co-operation with suppliers pays dividends.
• Develop single source supply by identifying and ameliorating risks.
• Suppliers should be responsible for the timely supply of good quality
products and they should bear the full costs of errors.
• Make life easier through quantitative assessment of suppliers.
• Equal partnerships reduce cost and give long-term success.
Note
1 Explained by Van Weele, A. 1994, Purchasing Management: Analysis Planning and
Practices, UK: Chapman and Hall.
10 Delivery quantities
• Repetitive supply.
• Types of orders.
• Calculating delivery quantities.
• Minimising work and inventory.
• Effective purchasing administration.
10.A Supply and suppliers

A large proportion of inventory control in a mature business will be the re-ordering
of supplies from an agreed vendor, and begins with a communication that a further
delivery is required. This situation is more about material control and scheduling
than traditional purchasing.
If the purchasers can separate out those items where there are routine orders
from those which are not likely to be repeated, the initial negotiations can focus on
ongoing delivery performance and presentation as well as quality, functionality
and price.
Where there is regular demand, lead time becomes less important and
consistency of delivery performance is key.
The strategy for supply should take into consideration the overall commercial,
logistic, inventory and added-value services which are to be provided by the supply
chain, and a variety of purchasing options can be considered, as appropriate. Supply
can be obtained from:
• a general supplier (internet, retail, general merchant);

• a dedicated provider (specialist or distributor);
• a manufacturer.
Delivery quantities 141
Attribute Manufacturer Open market stockist Specialist

Price x
Quality – consistency x x x
Quality – function x
Delivery speed x
Delivery reliability x
Lead time x x
Quantities x
Technical attributes1 x x
Development2 x
Convenience x x
Flexibility x
Range of supplies3 x x
Location x
Terms/finance
Trust/style x x
Figure 10.1 Supplier characteristics.

1
Provide R&D, marketing or complementary skills.
2
Innovation and new offerings.
3
Can provide products from a range of suppliers.
Each of these supply sources has different characteristics which should be exam-
ined to ensure their use is right for the items to which they are applied. The attributes
shown in Figure 10.1 are a gross generalisation, but indicate the general expecta-
tions of the supplier types.
Current practice is to set up a supply pipeline which goes through as few stages
as possible. However, the nearer the original source (manufacturer) of the item,
the less flexibility there is likely to be, the longer the lead times and the higher the
supply volumes.
10.B Organising repetitive supply

Supply has traditionally been based on the process of ‘ordering’ followed by
‘delivery’. In situations where items are bought occasionally, this is fine, but
where there are repetitive orders placed on the same suppliers, this process is both
tedious and inefficient.
Going back to basics, what a customer needs is reliable delivery, as required.
The ordering process is just a means of communicating demand level. Because
of the single sourcing and partnership arrangements discussed in Chapter 9,
other approaches to maintaining repetitive demand are preferable. For instance, in
the extreme situation of many demands per day, a kanban approach is very effec-
tive (see Chapter 6).
The procurement can then be considered in two distinct categories:
1 Order-based purchases for non-repeating items such as capital purchases and

one-off buying for customer special requirements. For these goods or services,
it is the sourcing that is important and so the involvement of ‘Purchasing’ is
necessary.
2 Delivery-based purchasing for repetitive orders, since lead time is much less
important than regularity and reliability of delivery. In this case, the role of
‘purchasing’ is to arrange contracts with supply partners, and leave the details
of the deliveries to those who need the supply.
10.C Order types

With the increased pressure for effective supply, simple processes of one-off pur-
chasing are largely irrelevant, except for capital purchases and non-repetitive
businesses (e.g. fashion garments). No longer does a buyer act as a dealer, seeking
a number of suppliers, negotiating price and delivery, and ordering one batch. This
is far too time-consuming and ineffective for the bulk of stock items.
Scheduling supplies is part of inventory management.
Ordering practices separate purchasing aspects from the re-ordering aspects.

Purchasing is strategic and carried out at intervals to set the structure, volume and
price of supply. Re-ordering is carried out as a result of stock control and forecasting
processes as necessary for managing replenishment. The two aspects require differ-
ent skills and can be carried out by different individuals. Supply can be broadly
classified into:
• one-off orders;
• bulk purchases;
• annual contracts;
• split-delivery purchases;
• rolling schedules;
• kanbans.
An analysis of these types of processes and their application is shown in Figure 10.2.
The more organised companies use JIT or scheduled supplies. Unfortunately, the
traditional software used in purchasing systems aims to limit customer commit-
ment, and so likes to have a fixed ‘order’ value. This means that rolling schedules
and kanbans often have to have a separate system. Schedulers are forced
to carry out extra work and make schedules look like split-delivery orders.
Supply agreement Advantages Risks Application

One-off orders No commitment No continuity of Projects and capital
supply purchase
Bulk purchases Traditional method High inventory C class items
for low prices High risk
Annual contracts Assured supply from Need to forecast Seasonal purchases
awkward supplier accurately one year (food)
ahead
Split-delivery Lower inventory Commitment varies Uses ‘order’-based
purchases risk at re-order time system
Rolling schedules Assured supply, low Non-adherence to A class items, delivery
inventory, low schedule, poor more frequent than
commitment forecast adjustment fortnightly
Just-in-time ‘Zero’ inventory pull Delivery failure High volume limited
system Unreliable forecast of range
Short effective lead demand Local good quality
time suppliers
Figure 10.2 Review of supply agreements.
Rolling schedules and kanbans are the only options which meet the real needs of
both the supplier and the customer.
10.D Order quantities

If the aim is to minimise overall cost by keeping order quantities low, then efficient
stock monitoring and goods received systems are required. If this necessitates
more frequent ordering, the purchasing workload is increased. However, an
increase in inventory actually results from the delivery, not from the ordering, and
the prime task must be to reduce the delivery quantity.
Control delivery quantities, not ordering.
Consider the ideal requirements of the supplier and the customer:
• The supplier wants an ongoing revenue stream of significant value to sustain

the business.
• The customers want regular, guaranteed flexible supply to give them agility to
meet their demand without holding stock.
• Neither wants poor quality.
The supply relationship is therefore about how to share the risk and cost. Before a
purchase order is placed, the supplier might have risk in keeping stock. When the
order is placed, then all the risk is transferred to the customer. Sharing the risk in
another way should be more beneficial for both parties. So, the supply chain works
better with contracts and schedules. If a business has an ‘order’ philosophy, then an
order with several delivery dates can achieve the result, although in a laborious way.
10.E Delivery quantities

Delivery quantities can be reduced in practice without incurring additional
distribution costs.
This can be achieved in two stages:
1 Increase the variety of items per delivery. This enables the delivery value to
be kept high, but the inventory of each line to be reduced.
2 Increase the delivery frequency. This enables the quantity delivered per line
to be reduced directly.
10.E.1 Variety per delivery

The stock value is decreased in direct proportion to the number of lines delivered
and the delivery quantity of each (reducing the lead time only reduces stock by the
square root of the lead time, which is usually very little).
Go for fewer orders and more frequent deliveries.
Consider the simplistic situation of a weekly delivery from one supplier who
delivers two products, one product one week and the other the alternate week.
Therefore, they deliver a fortnight’s worth of each item each time, which causes
an average of one week’s stock of each. Instead, it is then agreed that half the
quantity of both will be delivered every week. The delivery quantity of each is
now enough for one week and the average stock is then half a week’s worth.
The stock is halved by simply splitting the order quantity between the items
supplied. As paperwork could be increased by smaller deliveries, it is important to
have efficient recording and receiving systems in place to offset this. However, as the
main benefits are from A class items, this concept is usually only applied to them.
10.E.2 Delivery frequency

More frequent deliveries can give higher transport costs. If the costing is based on
weight or size, then, of course, load size makes no difference to the cost and quan-
tity reduction is easy (this is often the case for heavy or bulky goods). Where it is
considered that doubling the delivery frequency will double the costs, then more
resourceful solutions are required.
Options are:
• negotiating improved contract with distribution company for the higher volume;
• seeking alternative distribution process (groupage, courier, backloading, parcel
post, etc.);
• using a local supplier or agent;
• finding alternative suppliers with small lot mentality.
In many cases there is a cost compromise; a small additional cost of distribution

can lead to a significant reduction in warehousing and handling cost as well as
capital reduction on the part of the customer.
A class items should be scheduled or kanbanned.
10.E.3 Pareto-based delivery quantities

The key decisions in stock control are when to trigger replenishment (discussed
in Chapter 8) and how much to have delivered, because the larger the delivery
quantity, the greater the stockholding and obsolescence risk. Very small delivery
quantities may increase the workload, so it is best to restrict these to major cost
items – these are the A class items. Pareto Analysis (described in Chapter 3)
provides the basis for the best balance between usage rate, costs and operations
effort and cost.
The simple logic is to order A class items frequently to minimise stock value
and order C class infrequently to avoid too much work and enable more focus on
getting A class items right. Typically, the balance of delivery quantities is shown
in Figure 10.3. As C class items are the bulk of the inventory, this approach gives
less inventory and less work.
Get A class items delivered weekly or daily.
This, of course, is not dependent on the lead time taken by the supplier. Deliveries
can be arranged every day even if the lead time is four weeks, it just means that
there are about 20 supply orders outstanding all the time. The lead time does affect
safety stocks and therefore the overall stock level (it is a common fallacy that
order size should cover the supply lead time). The delivery quantity is only deter-
mined by the amount needed until the next delivery.
It is a gross assumption that a week’s worth of stock is the appropriate order
quantity for A class, but it appears to be generally correct across a wide range of
Pareto class Stock cover (weeks)

A 1
B 4
C 10
Other classes Customer demand quantity when needed
Figure 10.3 Pareto-based delivery quantities.

businesses and markets. The main exceptions are fast-moving goods where there
is daily delivery or kanban supply and the equivalent balance in days is used as the
order level. Companies where A items are delivered less frequently than fort-
nightly, have large stores and unbalanced stocks. Even when lead times are long,
this can be achieved.
Where usage rates are low, demand for A class may be too slow to support
weekly supply. In this case, the ratios 1:4:10 can be retained, but the ‘week’
changed to a longer timescale, just as it is shortened in fast-moving situations.
Pareto-based delivery quantities work well.
It is often safer to consider the current customer order book when calculating
supply quantities. In theory, any outstanding customer orders are part of the current
period demand, so do not have to be considered because they are covered by the
safety stock. However, in circumstances where there is an excessively high demand
(often a one-off demand that requires all the stock and creates back orders), it is
sensible to add this quantity onto the delivery amount. This will then get delivered
as an additional quantity one lead time ahead.
The Pareto order pattern fulfils the KPI requirements of minimising inventory
and workload. This is illustrated in Figure 10.4. If 1,000 are ordered monthly, this
will give 12,000 review decisions per year. Using the ABC scheduling pattern, the
number of decisions required is reduced to 11,440 (just under 5%), but the main
effect is that the overall investment in inventory is reduced from half a month
(about two weeks’ worth) to 1.325 weeks’ worth (plus safety stock in both cases).
The ABC order quantities reflect the natural ordering practice for many inventory
controllers, but the Pareto technique provides rules for controlling all stock in an
appropriate manner and makes it work better.
An approach to ABC type delivery quantities has been developed by Relph and
Milner.1 Their ‘K factor’ theory, although it is based on economic order quantity
(EOQ; see section 10.F.1), takes a very low ordering cost, producing results from
which delivery frequencies are concluded. Their analysis gives six classes instead
of the three ABC, and hence provides lower inventory investments.
Class Pareto moving Suggested order Pareto Average Total orders

lines (%) cover (weeks) turnover (%) weeks of value per year*
A 10 1 65 0.325 5,200
B 20 4 25 0.500 2,600
C 70 10 10 0.500 3,640
Total 100 100 1.325 11,440
Figure 10.4 Pareto order patterns for 1,000 stock lines.

* For 1,000 moving stock lines.
10.E.4 Horizon cover calculation
Instead of taking the current usage rate when assessing delivery quantities, it would
be more accurate to take the usage rate when the item gets delivered. This is not a
major problem because another order is triggered by the review level whether the
forecast is correct or not (in this section, the word ‘order’ is being used to denote
the quantity of goods over the period for which the customer is committing to sup-
ply, so it can be a firm schedule or, indeed, a single supply order).
The delivery quantity is dependent upon the:
• forecast usage rate;

• ABC class;
• delivery frequency policy.
Supported by knowledge of the lead time.

There are two different approaches to the calculation of delivery quantity:
1 Order quantity at lead time ahead (or fixed schedule length ahead).
2 Total requirement to the ‘order horizon’, less the amount already provisioned.
The order quantity calculation looks ahead one lead time and assesses the
demand over the order cover period at that time. In Figure 10.5, the order
quantity is the size of the hatched area (the order cover period is determined
by the ABC class). Schedule quantities or orders are required to be created for
this amount.
The order horizon approach looks at the total quantity required to the end of the
order cover (demand in lead time plus order cover) and deducts the amount already
provisioned (i.e. on order and already in stock). This is shown as the hatched area
in Figure 10.6.
Demand quantity
Expected
demand
Lead time
Order cover
Time ahead
Figure 10.5 Classical evaluation of order quantity.

Demand quantity
Expected
demand
In stock and
on order
Order
Lead time
cover
Time ahead
Figure 10.6 The quantity required by the horizon cover calculation.
Use horizon cover delivery quantities where demand is sporadic or seasonal.
The order cover calculation assumes that the previous forecast has satisfied the
requirement up to the lead time away, and extends the order cover by the amount
required in the cover period. If the demand level changes, then the next order trigger
will be earlier (for increased demand) or later (for decreasing demand).
The order horizon method builds the shortfall or excess into the order quantity
calculation. This is an advantage when using TSLs. The relative quantities calculated
are likely to be as shown in Figure 10.7.
10.F Scheduling supply

The best arrangement for ensuring reliable supply and low stocks is the rolling
schedule. It is commonly thought that parts which take a long time to arrive should
be ordered in big batches. This is not true. Large savings in inventory cost can be
made by receiving regular small amounts of the high turnover value items which
have long lead times.
For example, an imported part may take two months from the order to receipt into
stores. If there are 420 used per month, then an order could be placed every month
for about 420. This would lead to a stock of 210 on average (plus safety stock).
Alternatively, an order can be placed each week for delivery in two months’ time
from the date of each order. This would mean a delivery quantity of 100 each time
and average batch stock of 50. However, if it were possible to get delivery every day,
the batch stock would be reduced to an average of 10 (distribution methods have to
be altered to deal with more frequent transfers without incurring extra costs). This is
a significant reduction in stock and cost without any effect on customer service.
The number of deliveries per year may increase, but the workload can be reduced
instead by using supply schedules rather than individual orders.
Rolling schedules reserve supply and reduce risks.
Example
A valve wholesaler provides a 26-week full schedule to suppliers and
the first six weeks is a firm commitment. At the end of each week, the
schedule is rolled on automatically so that the new sixth week becomes
firm. No effort is required to do this as long as the requirement is the
same as it was a week ago (the forecast of the new twenty-sixth week
also needs creating at that time).
This successful application is then linked to the demand forecast so
the schedule outside the six firm weeks varies automatically with the
rise or fall in demand level.
Rolling schedules are not fixed on an annual basis because this can cause prob-
lems at the end of the contract where there is either a shortfall or a large unwanted
amount still to be delivered. The best schedule is an agreement between supplier
and customer:
• In the short term, supply quantities are fixed.

• Outside this period, the demand forecast is not binding but helps the supplier
to organise capacity and materials. Suppliers would prefer long fixed sched-
ules, while customers would like very short commitment. The scheduling
technique has many advantages when operated properly. It needs the full
commitment of both supplier and customer to keep to the schedule.
The benefits of scheduling supply include:
• focus on reliable delivery, not ordering;

• demand level is anticipated instead of created when an order is placed;
• supply lead time is not important (delivery quantity and frequency are);
• least risk for both customer and supplier;
• reduced stocks for customer and supplier;
• the opportunity for suppliers to plan their suppliers and future capacity;
• regular flow of materials;
• less purchasing effort;
• process improvement justified based on ongoing demand.
The reduction in purchasing effort results from the ‘rolling on’ of the schedule.
Not only do rolling schedules embody the principle of ‘no change – no action’
(the classic management by exception principle), they also improve the stability
of supply. Even if the supplier’s lead time varies, the scheduled quantity is not
affected, nor should the regularity of deliveries be affected either. A schedule is
essentially a sequence of delivery times, not order placement times, so the
equivalent of lead time for schedules is the inflexible front end of the schedule
(six weeks in the example). Where suppliers alter lead times, the schedule equiv-
alent is increasing the length of the fixed part of the schedule, so accuracy is
more important.
Rolling schedules are the fairest way to get supplies.
On a practical level, when the purchaser goes on holiday, there is no need to do

extra work, since the supply continues to arrive (unlike ‘orders’, where they may
have to be placed beforehand or else nothing arrives). Schedules solve many sup-
ply reliability problems and minimise risks.
For scheduling to work, there should be a simple agreement between supplier
and customer that the schedule quantity rolls forward unless there is specific
instruction otherwise, so that supplies are ensured. Both should be committed to
the quantities agreed. Any changes should be small percentage adjustments within
specified limits. Omitting scheduled deliveries is not countenanced, and major
changes outside the fixed period are avoided. Late or over-correction to demand or
stock levels, if not avoided, will give a stop–go requirement pattern. When usage
changes, the schedules should be modified gradually. Minor action early prevents
a major problem later on. The customer has to take some risk (or stock) when their
scheduling is wrong.
Accurate systems are required to control scheduled orders. Cumulative delivery
quantities should be correct and audited to ensure that scheduled deliveries are
actually being received. Linking the supplier into the customers’ systems enables
schedules to be updated automatically.
10.F.1. The ‘economic’ order quantity (EOQ) problem

Before the advent of lean supply and supply chains, there was a ‘batch and queue’
approach to supply. This led to a simple assessment of supply batch size in terms
of the cost of holding stock and the cost of ‘ordering’. This of course ignored the
more important costs that we consider today. For many decades, practitioners
knew that the EOQ calculation gave inflated inventory levels and the parameters
had to be ‘adjusted’ to give more sensible order quantities.
The formula is
A S
Order quantity, Q =
r P
A is the predicted annual demand, S is the cost of raising an order, r is the cost of
holding stock (finance and warehousing) as a percentage of the price, P, of each
item. Of course, this does not consider that much of the ‘ordering’ costs are fixed
and that effective ordering is now through contracts. The issues which should be
taken into account when ordering are:
• ABC class (not unit cost);

• batch quantity in other parts of the supply chain;
• batch quantities of complementary items (e.g. assembly components);
• handling and pack quantities;
• logistics and real warehousing costs;
• capacity constraints in the supply chain.
In the end, the smaller the better: it is what the supplier is willing to do for the
customer without the customer incurring extra cost. There appear to be no practi-
cal situations where it is sensible to use the EOQ formula.
EOQ cannot be used as the basis for sensible batch size evaluation.
As a simple critique of EOQ:
• A, annual demand, is a long-term forecast and therefore obsolescence risks

are inflated.
• S, the cost of raising an order, does not vary with batch size. It consists of
largely fixed personnel costs (halving the batch size therefore doubles S, so it
is not valid in this formula).
• r, the cost of holding stock, is largely a fixed warehousing cost, which is not a
percentage of the price.
• P, the price of each item, has been the subject of modifications to this formula
because of volume discounts.
Even if the linear relationships in the EOQ formula were correct, this line-by-
line approach does not apply in reality where lean supplies should be co-ordinated
across products, not only in manufacturing (material requirements planning;
MRP), but load-filling and lean balance.
EOQ was superseded by lean supply.

10.F.2 Minimum order quantities
There are situations where the supplier imposes a minimum batch quantity, or a
cost penalty, for ordering small quantities. This suggests that there is a mismatch
between the supplier’s perception of a ‘reasonable amount’ and that of the custom-
er’s requirements, and it is a sign that the supplier is not interested in the amount of
business that is available. Good liaison between purchaser and supplier could set up
a rolling schedule to assure the supplier of continued custom, albeit for small quan-
tities. The minimum could be an order value, in which case the purchaser can
explore more item variety as well as time-phased deliveries to meet this value.
Minimum order quantities, as set by suppliers, are generally arbitrary round
numbers. It is possible to erode them by a few per cent in many instances, and
gradually move to the required quantity.
Minimum order constraints on A class items should be studied energetically to
avoid expensive stockholding. Where the minima are imposed on C class, the
solution may be to accept the constraint as long as goods are likely to be used in
under a year. The minimum order quantity problem can be resolved in a number
of ways, offering different degrees of flexibility, as shown in Figure 10.7.
Where the supplier is a sole source, or branded goods, then more effort has to
be made to resolve the situation:
• The item can be sourced through a third party (avoiding significant cost increase).
• Customers may be persuaded to accept an alternative.
• A second source may be found.
• The item could be purchased to order and this minimum quantity supplied
directly to the customer.
It is important to avoid build-up of inventory in the supply chain. Negotiation

and understanding of the situation can sometimes lead to a sensible accommoda-
tion on both sides.
Minimum order quantities are a policy, rarely a technical constraint.
Constraint Initial policy Long-term changes

Minimum order value Place order for phased Negotiate on volume of annual
delivery business
Minimum shipment size Co-ordinate orders for Change method of shipment
widest mix of purchases
Minimum order quantity per Order for phased delivery Schedule, or arrange payment
line or alternative supply for when stock is used
standard item
Minimum delivery quantity Share item with others Renegotiate, or look for
per line (locations or competitors) alternative supply
Figure 10.7 Relative quantities.

10.G Supply co-ordination
Efficient supply chains hold little inventory. Managers should employ techniques
to ensure this happens. Examples of best practice are:
• For a single warehouse: arrange a schedule supply at the demand rate.

• For distributed stockholding (group or trade organisation): centralise purchasing.
• For standard products: use the same supplier or VMI.
• For distributed stockholding: co-ordinate supply to all stores, central distribution
centre with no stockholding on fast-moving products.
• For manufacture to order: buy and make the required quantity plus any loss
allowance.
• Parts for assemblies: make balanced sets, stock parts and only fast-moving
assemblies.
In manufacturing, batch quantities should be the same for purchase and produc-
tion where possible. The actual batch sizes are often determined by the practical
limits of capacity on a process, the overall process throughput time and the phys-
ical need to make a batch a handleable quantity. By co-ordinating the manufacture
of like products, the batch size of each can be reduced because the set-up costs are
reduced. Thus, a slow-moving item should be produced in small batches along
with similar fast-moving items.
10.H Purchasing processes

The stages of purchasing for a re-order are usually as follows:
1 Demand reduces inventory.

2 System is updated if required.
3 System identifies need to replenish.
4 Need is assessed and actioned if necessary.
5 Supply is initiated.
5a Authorisation is confirmed (if high value).
6 Requirement is sent to supplier.
7 Supplier initiates supply.
Taking out routine jobs gives time for better decision making.
These stages can be done efficiently and reliably through the operating system.
This is an area where continuous improvement gives good results. Consider the
process:
1 Demand reduces inventory: The whole process depends on the accurate and
immediate recording of usage quantity. Attention to this process is extremely
important.
2 System is updated if required: Where the system is not fully integrated, forecast-
ing and variability information have to be used to recalculate the review level.
3 System identifies need to replenish: Proper forecasting and inventory control
systems using the techniques described in this book will enable the system to
recommend replenishment (either through the review level or the safety stock
level – see Chapter 8). The information provided by the system should initiate
orders for a minimum of 85% of the items where orders are recommended.
If this is not the case, the stock control system is inadequate and requires
improving. The development of good stock control systems through forecast-
ing or MRP is not a complex computer problem. Solutions are available for
most situations, including sporadic demand and seasonal sales.
4 Need is assessed and actioned if necessary: In addition to the recommenda-
tions of the system, the inventory controller should pay attention to likely
events and vary the supply quantities as a result of feedback from customers
and marketing, although their predictions are notoriously optimistic. It is
always useful to have information from stores personnel, who should report
any items they consider to be at risk.
5 Supply is initiated: The actual purchasing process can be a simple confirmation
at the press of a computer key, or a time-consuming discussion with one or
more suppliers. For routine replenishment items, the ordering process should
be done effortlessly, once the order quantities have been determined (based on
ABC plus ‘events’). For C class items, in large-variety inventories it can be
practical to trust the package to order the correct amount without vetting.
Organise routines that minimise replenishment workload.
5a Authorisation is confirmed (if high value): For capital goods and non-routine
purchases, there has to be executive involvement in the commitment of company
money. For routine supplies, this process only serves to delay the purchase,
and is a great disadvantage. The signing of reams of purchase orders by a
senior person wastes their time and usually the orders are not investigated
during this process. There should therefore be generous spend limits for buyers.
Executives need to control purchases by:
• Monitoring the value of the order placed and the outstanding order com-
mitment (weekly or monthly).
• Signing the major value contracts (say 20 per week).
• Auditing the general order level by random checks each week.
• Setting targets for reduction of value of purchases relative to demand
(but without sacrificing quality or delivery programmes).
6 Requirement is sent to supplier: Normally, this can be done over the internet.
Email is not ideal as it is a text-based system, and orders need reading into the
suppliers’ systems, entailing additional work. Liaison between supplier and
customer can ensure that this process is accurate and easy. Companies should
ensure that there is a proper up-to-date list of authorised purchasers for each
supplier and that the formal process is not by-passed ‘in the interests of quick
delivery’ because the system is too slow or cumbersome.
Where suppliers need a formal contract or signature, then post, authorised
media or even fax can be fast and efficient, as long as the processes are
designed carefully.
There is a legal aspect to procurement that must be observed and which
occasionally needs to be enforced when there is a dispute. Schedules and
orders should therefore be created and communicated within this framework.
The well organised purchaser uses the phone or interactive media (e.g. Skype)
for regular supplier liaison and negotiation, but will not have to use it for normal
routine ordering. The telephone is time-consuming.
Manual orders are to be avoided since they complicate the system. How-
ever, there is a good argument for a petty cash purchase system where over-
the-counter purchases of low value are carried out by an authorised person
outside the normal purchasing system. This will avoid a large bulk of formal
purchases and give immediate access to the items required (often mainte-
nance parts). Control over these purchases should be through the monitoring
of monthly purchase values.
7 Supplier initiates supply: The information provided for the supplier should be in
a form that is simple for the supplier to process into their system. Any variations
from the anticipated delivery schedule should be relayed to the customer without
them having to enquire. In situations where the customer is apt to alter the demand,
there is a tendency not to report difficulties. However, by keeping the customer
informed, programmes can be modified and consequences avoided in advance.
Once the demand has been accepted by the supplier, the delivery part of the
process is initiated, using the information provided by the purchasing process.
This will lead to delivery exactly as the customer has requested. If there are occa-
sions where this is not the case, investigation and remedial action should follow.
Key points
• Focus on delivery quantity and timing, not ‘ordering’.
• Organise supply to minimise delivery quantities and effort.
• Schedule supply for A class repetitive demand items.
• Use Pareto ABC as a basis.
• Apply lean supply concepts – definitely not EOQ.
• Structure supply quantities to reduce workload and inventory.
• Review purchasing control and administration to reduce work.
Note
1 Relph, G. and Milner, C. 2015, Advanced Methods for Managing Inventory within Business
Systems, UK: Kogan Page Ltd.
11 Forecasting demand
• The role and types of forecasting.

• Forecasting in the product life cycle.
• Analysing demand.
• Avoiding forecast data error.
• Benefits of group forecasting.
11.A Options for assessing demand

Businesses make large investments in inventory. Much of it could be avoided if
there were better forecasts of demand. This is relatively inexpensive, but requires
understanding, communication and commitment. Although people say, ‘forecasts
are always wrong’, the smaller the inaccuracy the less stock is required. Each time
goods are bought, it is a prediction that these items will eventually be sold.
Where forecasts are excellent, suppliers can predict when customers want the
items, so they can be obtained and delivered just when needed (e.g. automotive
assembly). Inventory managers need forecasts that identify the requirement of
each product line. Personal intuition (guessing) may work sometimes, but tech-
niques and communication improve the overall forecast quality and the stockholding
(increasing availability and reducing inventory).
Good forecasting means low stock. Poor forecasting means high stock.
Forecasting should be based on information which is accurate and appropriate

for the purpose. Where there is reliable historical information, this usually gives
good forecasts, but future sales may not be reflected by history, so market infor-
mation has to be used to identify potential changes in demand pattern. Unfortunately,
this information is not so reliable.
Clarifying the demand types will assist in forecasting. An analysis is:
• Continuing demand: Future demand continues from historical demand, so

historical data is excellent for forecasting, even when demand is seasonal.
Forecasting demand 157
• Scheduled requirements: Customers provide extended forward demand rate
information, so the need for stockholding is minimised and treated as ‘depen-
dent demand’ (see Chapter 14).
• Changes in conditions: Marketing input is necessary where future demand
levels will change significantly (through gain or loss of significant customers
or markets). The forecast is again based on historical data.
• Events: Service for one-off demands are completely dependent on estimates
from marketing experts. These can take the form of:
product promotions and offers;
focused advertising or sales campaigns;
large single customer orders (where the demand is more than four MADs
above the average);
replenishment resulting from product recall or modification.
Detailed analysis of demand, both historical and anticipated, provides better

forecasts. The better the understanding of demand, the more accurate the forecast,
and the more sophisticated the forecasting tool used to support this.
Forecasting techniques can be analysed in many ways. One simple classifica-
tion is between forecasts which use prediction and those which use history, another
is between qualitative and quantitative figures, and another between intrinsic fore-
casting and extrinsic forecasting.
Considering these concepts in turn, historical forecasts rely on looking back to
predict the future. This is accurate as long as there is continuity or gradual change.
Historical forecasting forms the basis of most inventory forecasting because it is
quantitative and has specific detail. It works successfully for most items, but obvi-
ously it will lead to disaster when there are unexpected events or where conditions
change rapidly (i.e. where predictive forecasting plays a role).
People will make predictions based on understanding of future events, but these
predictions are normally qualitative. A sales campaign will increase demand, but
by how much? Although the qualitative prediction is marginally useful, stock con-
trollers have to provision a specific quantity to fulfil the demand. Who is likely to
get the blame if it is wrong?
Good forecasting costs less than high stocks.
Much successful forecasting is based on a company’s own intrinsic data. Taking

into account external (extrinsic) information can enhance the forecasts. Extrinsic
information from trade statistics or market information or government organisations
are usually less specific, but enable demand profiles to be understood better. It is
convenient to split extrinsic information into leading indicators, concurrent indicators
and lagging indicators, according to whether the information collected shows future,
current or past activity. For instance, government statistics on housing starts is likely
to be a useful leading indicator to a national roof tile manufacturer. The same
information would be a concurrent indicator, showing what is happening to the gen-
eral demand this month, to companies hiring digging equipment for house founda-
tions and, for architects, it would explain the level of activity which they had already
experienced. Obviously, leading indicators are the most useful, especially if they
show demand further along the supply chain or activity which causes usage (e.g.
political unrest in one country causes increased demand from tourists in others).
The manipulation of forecasts is part of the inventory manager’s art, since com-
puterised systems are unable to respond to market information in advance, unless
primed.
11.A.1 Types of inventory forecasts

Some types of forecasting are used only for specialised purposes. Of those used
regularly by inventory managers, the following five approaches are the most used
for assessing requirements.
11.A.1.1 Market research

There is nothing like knowing for sure what customers are going to buy. Firm
schedules are the ideal. Sometimes it is possible to obtain firm order commitment,
but more often customers will only give estimates. Good liaison with the biggest
customers, or the biggest users of A class items, can have a major impact on fore-
casting accuracy. If customers trust a supplier, they will provide accurate informa-
tion, since they should be in the best position to know their market and their likely
order requirements.
An alternative would be to carry out market surveys to discover trends. These
can be particularly useful in investigations for new products or for determining the
effect of different sales or promotional policies. Devising a market survey to give
factual results for calculating demand is a skilled job. As an ongoing exercise, it is
only cost effective for a narrow range of high volume items.
Another method of assessing customer requirements is to ask the sales people.
The risk with this is that sales staff sometimes confuse the difference between
maximum potential sales and expected average sales. Consequently, their esti-
mates often have to be interpreted before use.
11.A.1.2 Market demand models

These are generally based on knowledge of the customer’s markets. If the major
factors which cause the demand can be identified, then a model can be created.
The influence of each factor can be assessed and put into an equation, or the
factors can be input into a simulation which can develop its own parameters.
Choose the right forecasting technique for individual products.

Models are very effective where there are leading indicators, and where there are
simple deterministic relationships causing demand. Some models can be financial
(expenditure on promotion versus increased turnover), others technical (item needs
replacing every x years) or based on commercial factors.
11.A.1.3 Historical techniques

Forecasts should be based on historical data, and then modified by the other influ-
ences. The only exceptions to this rule are for new products, and those which
have a causal link to another forecast (via a model or material requirements plan-
ning (MRP)). Historical techniques are therefore used for almost all demand
forecasting.
There is a wide variety of historical techniques to choose from. Some are
simple and give adequate results, others are much more sophisticated and give
excellent results in the right circumstances. In general, the more sophisticated
the technique, the more data is required to realise its potential, so established
products can be forecast best using these. Recent demand data is normally the
most important, so weighted averages are better than simple averages. The most
common basic approaches to historical forecasting are discussed in the following
two chapters.
Overall, the approach provides the most accurate answer for most businesses.
It requires thorough communication, knowledge and management, and management
have the challenge of identifying those items where it is grossly inaccurate.
11.A.1.4 ‘Maximum–minimum’ stock systems

Old or simplistic thinking or systems have these built-in parameters. The supply
quantity is the difference between the maximum and the minimum, and the mini-
mum is not the minimum stock – it has to be the review level. Historically, this
approach led to inflated stock levels since stock-outs caused low ‘minimums’ to be
increased while excess stock did not cause them to be reduced.
With the availability of spreadsheets, the formulae provided in previous chapters
can be used to control even the largest inventories, so there is no excuse for anyone
to use this concept.
11.A.1.5 Demand analysis

The demand for stock results from a variety of customer requirements, so instead
of using historical analysis (the intrinsic approach), investigate the causes of
demand (the extrinsic approach). This is particularly useful where there are a
small number of customers taking a large value of items. If the reasons for the
demand pattern are understood, inventory can be optimised to meet and improve
the service. By sharing information with the customer, inventory can be reduced
and availability improved; in the case of regular requirements, this could lead to
dependent demand, schedules or local kanban supply.
Special arrangements with customers are sometimes used for some high value
and critical items, or simply enhancing the historical system by better communi-
cation. This can result in reserving stock, thus increasing the inventory value,
whereas including planned demand from these customers will manage the stock
more efficiently.
11.A.2 Product life cycle

The market is continuously changing, mainly as a result of technology, fashion,
values and legislation. Products therefore have a life cycle which can be simply
divided into three phases: introduction, supply and obsolescence (Figure 11.1).
Both marketing and design activities have more complex cycles, which start
earlier.
It should be noted that for many products, life cycles are getting much shorter
as technology develops and social media speed communication.
Historically, the product life cycle for a mechanical typewriter could be

25 years, for word processors (what were they?) it was about two years, a
laptop was a few months, and now it is daily upgrades on tablets and
phones.
For a service organisation, the need for stockholding extends beyond the prod-
uct manufacturing cycle since customers will retain products for many years after
the end of manufacture and still expect spares to be available to support a service
for their equipment. For some companies, products manufactured more than
20 years ago are still in use and being repaired as required. Some service organi-
sations have announced stop dates for support in conjunction with sales initiatives
to sell customers new products (e.g. software suppliers). Elsewhere this frequently
fails because third party maintenance organisations seize the opportunity to take
over this business, which can be very lucrative.
Average demand
Introduction Supply phase Obsolescence

phase phase
Time
Figure 11.1 Product life cycle.

11.A.2.1 Introductory phase
This phase should follow the design and development phases. Prior to this, the inven-
tory planner will be working with the product project team, when the marketing and
customer service strategy for the product should be decided. To determine the stock
requirement, the inventory planner will want to know the forecast demand and the
possible geographical distribution of customers, which will enable the planner to
determine the requirement and place time-phased orders for the items necessary to
support the marketing strategy for the product ahead of the launch date.
Product launch is a critical period in the life of the item and it is essential to have
sufficient to support the first customers. However, if too much is bought initially, the
profit can be seriously impacted because much of the initial purchase may never be
used if too much ‘just in case’ inventory is bought. The planner must monitor care-
fully early life usage as the sales build up, and adjust ordering accordingly. A simple
approach to limiting the risk at this introduction stage is shown in Figure 11.2.
There should be two forecasts created: the optimistic one (usually by the mar-
keting department) and the pessimistic one (usually for the accounting revenue
budgeted). The initial inventory needs to cover the optimistic forecast just in case
it is correct, so the minimum initial inventory should be enough to cover the opti-
mistic forecast during the supply lead time. However, it would be unwise to have
more than required until the item is superseded, so this limits the initial inventory
holding to the projected pessimistic forecast quantity over the time until the prod-
uct would be terminated. As the product begins to sell, the forecasts converge and
inventory can be gauged better. Careful monitoring at this time is essential.
Ideally, the inventory planner would like new products to be launched gradually
to the market so that the inventory does not have to be so high and the risks of
stock-out and obsolescence are reduced.
Optimistic forecast
Forecast demand >>>
This area is the

minimum safe
stockholding Pessimistic forecast
This area is the maximum safe

stockholding
Lead Forecast
Time >>>
time horizon
Figure 11.2 Forecasting new products.

11.A.2.2 Supply phase
Established products are controlled in the normal way by the inventory techniques
described previously and the forecasting techniques in the next chapters. Inventory
control is concerned at this time with maintaining availability.
11.A.2.3 Obsolescence phase

Old product lines are not of interest to sales and marketing, but are very important to
inventory planners. Poor control here is a major cause of excess and obsolete stock.
The task now is to prevent a build-up of surplus stock as demand declines. The first
step is to identify on the system that the item is in the obsolescence phase, so the
stocks diminish toward zero at the agreed termination date (in practice, the termina-
tion date often slips due to optimism from technical and marketing departments).
At the start of the obsolescence phase, there may be significant inventory and the
inventory planner has to plot carefully the inventory profile in order to maintain item
availability until the line is withdrawn. Consideration should be given to the relative
importance of customer support and the high risk of having to write off unsold stock.
When it becomes obvious that inventory will become surplus, the techniques
described in Chapter 4 should be employed, utilising the process in Figure 4.4,
particularly taking any opportunity to redeploy inventory in sister companies
where the item is not its obsolescence phase.
As any obsolete items must be financially written down to zero value, management
has to make a decision whether the nuisance and cost of keeping items outweighs the
risk that at some time in the future they will be required, and difficult to source.
11.B Causes of forecasting inaccuracy

People sometimes say, ‘you can’t forecast this’, a statement which is untrue. Every-
thing is capable of being forecasted, but some things can be forecast well and others’
forecasts may be very inaccurate. What causes poor forecasts? There are a variety
of process and operational causes as well as personal agendas, including:
• inaccurate data;
• sales information rather than demand statistics;
• bias;
• speed of response to change;
• poor assessment of supply capability;
• inclusion of extra demand in the forecast;
• shortage of data.
What can be done about these?
Get the data right – then you can forecast.

11.B.1 Inaccurate data
Inventory management is based on accurate stock, supply and demand data, so
that appropriate controls are maintained and actions taken. Inaccurate data leads
to poor availability, unreliable delivery promises and excess stocks. It is therefore
imperative for stores to maintain records accurately in real time, and not end-of-
shift or weekly physical stock checks. Techniques for improving records are:
• stores empowerment (personnel identify stock-outs and potential problems);

• automated recording and checking systems (bar coding, radio-frequency
identification (RFID), in-line counters, weigh counters);
• batch control (leave supply batch intact and agree record quantity when
opened);
• perpetual inventory checking (continuous audit to find discrepancies and
eliminate causes).1
Where records are inaccurate, improving these should be the first priority so
that the inventory management calculations can be relied upon.
11.B.2 Sales information rather than demand statistics

When considering inventory statistics, it is normal to discuss ‘demand’. However,
most companies record ‘sales’. The difference is only important where the actual
amount issued differs from the amount ordered. This could happen when rounded
box quantities are supplied, or shortages make suppliers allocate proportionally or
there is a stock-out.
Where there are repetitive stock-outs, the lack of sales could lead to the conclu-
sion that no stock is required, unless demand is also recorded. Professional stock
records should therefore monitor ‘demand’ (order requirements at their delivery
dates) rather than just sales.
Example
An item has a demand in three months of 32, 48 and 40, but there is no
stock in the first two months and then a delivery of 90 in month 3.
During the next month, the sales records would give monthly sales of 0,
0 and 90. The actual demand is 40, with MAD of 8, whereas the sales
statistics would give an average of 30 and MAD 40, resulting in under-
forecasting and higher safety stocks.
11.B.3 Bias
Sales tend to use a forecast either as a motivator to the sales team by exaggerated
optimism, or to ensure that stock is available. Consequently, projections from
sales departments are notoriously high. However, in some organisations sales
forecasts are low so that when they are exceeded, bonuses are distributed. Stock
controllers should understand this and interpret sales data for forecasting. If man-
agement will make forecast accuracy a KPI for marketing and sales departments,
and charge consequential surplus inventory to their budgets, the quality of fore-
casts would improve.
11.B.4 Speed of response to change

It is difficult to get the right compromise between over-reaction to events and lack
of response. One or two periods of high demand could signify an increase in market
demand or simply a statistical variation. Either assumption will cause risk. Similarly,
when considering what time intervals to use for aggregating data, if the time
period is long (e.g. a month), there could have been major changes in the customer
demand profile within that time.
11.B.5 Poor assessment of supply capability

Fulfilment of increasing demand may be restricted by suppliers’ capabilities. Even
good suppliers with increasing volumes become overloaded and cease to supply
successfully. Suppliers can be optimistic about their ability to supply, since they
will want to provide other customers also. Good liaison and planning can identify
risks and sales diverted to avoid problems. This is a particular issue with short-
term peak demand such as sales promotions.
11.B.6 Firm scheduled orders

Customer schedules can appear to add variation to sales. However, it is not ran-
dom demand and does not need forecasting. It can be assessed separately and
makes forecasting easier.
Example
The weekly demand pattern is:
20 510 30 5 25 520 15 35 25 515
There is a major customer who requires 500 every four weeks and a
variety of customers who take 15 per week on average. A sensible policy
is to stock for the many small demands and to order in specially for
the 500 to arrive just before it is required. This is negotiated with the
customers and separated in the stock record.
11.B.7 Events
Special offers, campaigns and promotions are in the demand data, but need to be
extracted when creating sensible forecasts. It is debatable how to deal with histori-
cal forecasting during the event period. A preferred way is not to include the event
periods in the calculations. This is valid unless there is fast change of underlying
demand rate or little available data anyway.
One-off events can be excluded from forecasts automatically using a filter set at,
say, four MADs. Demands outside this range are treated as non-stock sales.
11.B.8 Shortage of data

For a ‘good’ historical forecast, sufficient data is required. It is best to include all
data as far back as the demand is consistent with the current conditions. The
obvious risks of using old information are:
• undetected population changes have altered the usage rates;

• demand patterns have changed through modifications in user practice;
• some requirements have been modified or substituted by alternatives;
• contracts or customer base have changed significantly.
Forecasting models can often cope with poor old data.

For seasonal sales, it is essential to have data for more than one year. Forecasts
are poor when there is a small amount of information available (Figure 11.3).
Example
Recording ‘the last two months’ sales were 500’ gives very little informa-
tion on which to base stock levels. If the record says ‘sales were 300 in
month 1 and 200 in month 2’, then this starts to suggest a likely demand
pattern. If weekly demand data were available, then it would be possible
to see whether the demand was:
• only two customer orders or sporadic demand;

• regular orders each week or each day;
• decreasing demand.
It is better to use more, smaller time periods rather than large ones, unless the
demands in the time periods become very inconsistent or non-standard. For exam-
ple, if demands on Fridays are different from the rest of the week, daily time peri-
ods would give cyclic demand with a peak each Friday. As long as the forecast
deals with this, the daily data is better.
60
40
Mean error as % of MAD
20
0
0 5 10 15 20 25 30 35
20
40
60
Sample size
Figure 11.3 Effect of amount of history on accuracy of forecast.
How much data is needed to make a forecast? Figure 11.3 shows how the accur-
acy improves with more data, and more than 12 periods of data provides reasonable
accuracy. The size of the forecast error depends on how erratic the demand is – as
indicated by the vertical scale. There is a reasonable degree of accuracy afforded by
12 periods of demand history and it starts getting good over 30. As the accuracy also
depends on the variability (MAD), more history is required for variable demand
products than consistent ones, as would be expected.
Work with the data available – it’s the best you have!
11.C Methods of improving forecasting

There are some simple ways for making forecasts most effective, which include:
• Update the forecast at each period end (ensure that data is up-to-date).
• Use improved historical forecasts (see Chapters 12 and 13).
• Get a responsible forecast from marketing, especially for events.
• Concentrate effort on A class lines (the systems can do C – understand A
demand).
• Ensure inventory data is accurate.
• Smooth demand (time-phased customer delivery schedules).
These can be implemented by any business.

11.C.1 Feedback and monitoring
Where forecasts diverge significantly from sales, the causes should be identified
so that the situation can be avoided in the future. This is manifest by large forecast
errors or stock-outs.
11.C.2 Data aggregation

Statistically, there is a major benefit from using as much data as possible (Figure 11.3).
As well as going back farther in time, data can be gathered from a wider geographical
area or additional similar businesses (which decreases the MAD).
If a wider geographic population is used, the risks are:
• Usage patterns differ regionally.

• Different demand patterns (or modification mix) exist locally.
• Data from different sources may not be compatible.
It is therefore a matter of assessment and balance which data should be used for
forecasting. In general, the wider the information base, the more reliable the forecast.
The accuracy improves as the square root of the amount of data included.
11.C.3 Comparative items

For high usage items, random fluctuations are small compared to the forecast demand.
Where demand is small, fluctuations can be large compared to the forecast.
By grouping several items together, the statistics become better and the overall
MAD decreases compared to the total demand rate.
If there are a large number of items in a product range with low usages, com-
parison with the major items can be a very good way of allowing for changes in
the underlying level of demand. The forecast for each item can be based on indi-
vidual historical demand, and the resultant forecast multiplied by the relative
change factor used on the major item.
Example
Consider two items in the same product range, one with a demand of
100 per month and the other with a demand of two per month. The
forecast for the major item shows that the forecast has drifted from 100
per month to 110 per month. Assuming that the profiles are similar, the
demand for the minor item should increase by 10% as well – to 2.2. It is
much more difficult to detect confidently a change from 2 to 2.2 than
from 100 to 110. It does not matter whether the two items change in
exactly the same way. This technique works if the minor item is more
likely to follow the major item than to remain static.
Key points
• Better forecasting gives low stock levels.
• There are many techniques and options, so several should be used.
• Forecasts can be predictive, based on product or sales information,
or historical data.
• Record accuracy is paramount for realistic forecasting.
• Understanding the demand improves forecasting.
• The more data, the better the forecast.
Note
1 See Wild, T. 2004, Improving Inventory Record Accuracy: Getting Your Stock Informa-
tion Right, UK: Elsevier.
12 Historical forecasting
techniques
• Historical averages.
• Better forecasts with weighted averages.
• Choosing the weighting factors.
• Choosing the best forecast.
• Warning when forecasts are bad.
12.A Basic forecasting techniques

Historical forecasting can deal with most situations for the majority of items.
There is no need to forecast for customer scheduled requirements unless the sup-
ply lead time exceeds the firm planned schedules. On the other extreme, where
there are ‘events’ where demand characteristics suddenly change, some warning
from the market is needed, whether this change is short term (such as a product
promotion) or long term (such as a gain or loss of a major customer).
A good forecast has to be:
• accurate;
• consistent.
Where demand is random, it will not be exactly correct so safety stock will be required
to cope with this. When forecasting is good, safety stock is small. The amount of stock
investment depends largely on the quality of the forecasts and the supplier delivery
scheduling.
12.A.1 Events and market changes

As discussed in Chapter 11, assessing demand is easy if the business knows what
customers require. In practice this requires the knowledge of customers, products
and background conditions. That comes from discussion and evaluation with users
farther down the supply chain and through techniques such as market research,
questionnaires and customer surveys. The strength of these techniques lies in
170 Historical forecasting techniques
providing general information on the levels of business and on the likely uptake of
new products, or the phase-out of old ones. This information is invaluable where
large changes in demand level are likely to occur.
These techniques need to be better managed to make them more acceptable as
forecasting tools for each line item, and inventory management need to work with
sales departments to ensure that they do not draw biased (usually optimistic) con-
clusions which then have to be second guessed for planning stocks. However,
assessment of the market demand is an essential tool for the professional stock
controller and the technique should be used for A class items and those at the start
and end of the product life cycles.
Some companies have leading indicators which help to assess potential sales.
They provide estimates, enquiry responses or have contracts and formal agreements.
These can be invaluable in estimating trends or requirements for individual product
lines. Of course, the length and variety on the customer order book is a good predic-
tion, although in some markets there is always the possibility that dates or quantities
will be changed.
Forecasting is therefore one of the important aspects of stock control. It is a
complex art which embraces many factors.
12.A.2 Demand history

Since the information from the market is normally scarce and imprecise, compa-
nies rely on history for prediction. For the majority of items this works well (i.e.
except for events). Historical forecasting has the advantages that the data is:
• readily available;
• detailed by item number;
• generally reliable;
• easy to use.
However, using demand history is rather like rowing a boat. To go forward means
sitting looking backwards and working hard. This is fine if there is no obstruction
in the way, but it is advisable to look round from time to time to see what is in front
(‘events’). In a similar manner, inventory controllers use history to determine the
future. They have the problem that they are steering many hundreds of items at the
same time, and it is not practical to look round for each one to see if it is heading
for collision. What is required is a general warning mechanism to detect impend-
ing disaster, and a forward look for the more sensitive and important items.
Forecasts should be based on history.
Historical forecasting works well for the vast majority of items and is therefore
a basic tool of inventory control. More sophisticated historical forecasts give better
results, but there is always the need to keep abreast of real changes in the market.
Historical forecasting techniques 171
In practice, a variety of simplified models, based on major features of the
demand pattern, can be used to meet stock control requirements for standard
items. Historical forecasting methods are based on the mathematical manipulation
of historical data. This produces excellent forecasts where the demand pattern is
consistent and has the advantage that it is:
• consistent;
• rapidly calculated;
• frequently updated automatically;
• based on facts, not opinion.
However, as it is a simple ‘model’, the inventory manager may have additional

information that will modify the requirement and can be used for important items
or groups of items.
12.B Moving average

The most straightforward method of forecasting the next period’s sales is to take
the average of sales for each preceding period, add them up and divide by the
number of periods (normally weeks). Moving averages take a fixed length of his-
tory (N periods) and when the average is recalculated at the next period end, the
oldest information is discarded and this retains the average over the last N periods
(see Figure 12.1).
Total demand in N periods

Moving average =
N
(c) Increasing
(a) Smooth demand (b) Variable demand
demand
200 200 200
Demand
180 180 180
6 week average
160 160 160 12 week average
26 week average
140 140 140
120 120 120
100 100 100
80 80 80
60 60 60
40 40 40
20 20 20
0 0 0
1 3 5 7 9 11 1 2 3 4 5 6 7 8 9 10 11 1 3 5 7 9 11
Figure 12.1 Moving averages.

Financial accounts have a fixed annual cycle in which history builds up from
the start of the year. This data is unsuitable for stock control, since a fixed number
of periods of history, N, is always used.
The selection of N is of prime importance in getting the best from the moving
average technique. It is a compromise between the inaccuracy caused either by too
little data (see Figure 11.3) or by using old, irrelevant data. Often, the average is
taken over 12 months (or 52 weeks) to avoid complications due to holidays or
seasonality. However, it has to be remembered that by averaging 52 weeks of his-
tory, the data used is, on average, 26 weeks old; a lot may have changed since then.
Seasonal forecasting is discussed in Chapter 13, but in this chapter it is assumed
that the demand is not significantly seasonal.
In general, a ‘suitable’ number of periods, N, is chosen for a whole range of
stock items. This is not the best option, but a practical compromise. Calculations
including many time periods are best where underlying demand is static or demand
fluctuates widely (Figure 12.1b). However, it is better to take a few recent periods
when demand is increasing or decreasing since this enables the average (forecast)
to respond faster to changing demand levels (Figure 12.1c).
Theoretically, each item needs its own averaging technique.
Taking an average is really a method of segregating trends and random fluctuations.

Where the average includes many periods of historical information, any spurious
effects are damped out, but this will also make the result insensitive to trends.
Conversely, a short moving average, with small N, gives good response to the
market but is affected by fluctuations.
The historical data shown in Figure 12.1 are compared with the 6-, 12- and
26-week moving averages that were created at the time of each period end. The
26-week forecast is the smoothest but in Figure 12.1c it is worst at following the
trend. With this trend, the 6-week average is most accurate, but Figure 12.1b gives
inconsistent forecasts with variable demand. Assessing the best individual fore-
cast is possible using focused forecasting. In general, more advanced forecasting
methods are used professionally except where demand is very variable.
12.B.1 Alternative calculation methods for moving averages

Updating moving averages by summing period-by-period demand can be achieved
by manipulating the formula. The new average can be calculated by deleting the
oldest and adding the newest; that is,
Total Demand + New Demand Oldest Demand

New Average =
N
Or even easier, and more practical, is to find the difference between the old
demand data (to be discarded) and the new period data, and divide the difference
by the number of periods in the moving average. Then this can be added to the old
average to give the new average; that is,
New period demand − Oldest period demand

New Average = Old Average +
N
Moving averages for variable demand C class items.
Moving average is most successful where the demand fluctuates widely, because
more sophisticated methods cannot easily identify demand trends and profiles in
the presence of sporadic demand. Moving average is often used for C class items
because the demand for these is often very variable compared to the mean, and it
is more important to have a forecast which is generally reliable rather than one
which requires some management.
Example
Consider that the demand for an item over the previous six periods was:
Period 1 2 3 4 5 6 7
Demand 17 11 15 12 14 15 ?
The average demand, Ao, is 14 for the first six periods (total demand was
84 over the six periods). This is the forecast for period 7, but the actual
sales were 11. The new six-period average demand AN is therefore 13.
To calculate the new moving average, A7, the demand for demand in
period 1 is discarded since it is older than six months and the new
data,11, is added (= 78). A simple way of calculating that A7 is 78 was
84 – 17 + 11 = 78. The forecast is 78/6 = 13.
Alternatively, the new forecast = 14 + (11 – 17)/6 = 13.
Moving average is a commonly used forecasting method, especially by ama-

teurs, because it is simple to apply and easy to understand. It has the disadvantage
that it assumes old data is as important as recent data. In most circumstances,
recent data is the most reliable for forecasting (seasonal sales are discussed in the
next chapter). A focus on the recent demand history balanced by older informa-
tion would be the best compromise.
12.C Exponentially weighted averages

As discussed, the two difficulties with moving averages are that it does not
respond well when requirement levels change and it gives equal importance to all
periods’ demands.
0.4
Alpha = 0.4
Contributions to the average 0.35 Alpha = 0.3
Alpha = 0.2
0.3
Alpha = 0.1
0.25
0.2
0.15
0.1
0.05
0
1 3 5 7 9 11 13 15
Age of information
Figure 12.2 Contribution from each period to the average.
An average which takes more notice of recent history and less notice of older
data has major benefits. The use of a weighted average improves historical fore-
casting. There are many options for weighting the averages, but one option appears
to be better than the others and is used almost exclusively.
Exponential smoothing avoids both problems outlined above. It produces a
weighted average which is based on all the information available (see Figure 12.2).
The calculation of a new average only requires the old average, the new demand
and a weighting factor.
The simplest professional forecasting technique is exponential smoothing.
The name exponential smoothing comes from the fact that contributions from
history form an exponential curve go backwards in time. All the historical infor-
mation available can be used, but the data gradually becomes less significant as it
gets older.
Forecasting with exponential weighting is the simplest of the professional fore-
casting tools. It adjusts the forecast for the last period using the actual demand by
mixing a portion of the old average (forecast) with a portion of the new demand, so:
New Forecast = α × Demand in last period + (1–α) × Forecast last period
The mixing portion, α, is the smoothing factor. Presented mathematically, this is:
A = αD + (1 – αΒ)
where: A is the new forecast (the new exponential average);

B is the old forecast;
D is the demand in the period just completed;
α is the smoothing factor.
The new forecast is calculated as soon as the data for the previous period
(usually week or month) has been collected. The new forecast is then used for
the period starting at that time. The value of α is between 0.1 and 0.4, as dis-
cussed below.
The smoothing effect of exponential smoothing is set out in Figure 12.3 and
illustrated in Figure 12.4. Like moving averages, this calculation simply analyses
the demand into average and an error which is treated as statistical fluctuation.
The smoothing factor, α, defines the amount of importance that is given to each
period’s demand.
In Figure 12.3, the next period’s sales were predicted each period using exponen-
tial smoothing with α factors of 0.1, 0.2 and 0.4. The forecast for the next month
was calculated using the formula and from ten periods before, finally providing a
final forecast for the current period (11) (see also Figure 12.4).
12.C.1 Selection of the smoothing constant, α

The sensitivity of the forecast can be changed by altering the value of the smooth-
ing constant, α. Low values of α make the forecast consistent, whereas high values
make it reactive to change. Making α too high (above 0.4) would render the result
Period Demand Forecast demand

α = 0.1 α = 0.2 α = 0.4
1 57 99 113 99
2 35 95 102 85
3 171 89 88 73
4 49 97 105 129
5 62 92 94 89
6 189 89 87 80
7 109 99 108 120
8 36 100 108 100
9 30 94 94 76
10 131 87 81 74
11 178 92 91 100
12 Forecast 100 108 127
Figure 12.3 Exponential forecasting sensitivity.

(a) Smooth (b) Variable (c) Increasing

demand demand demand
200
Demand
180 Alpha = 0.4
Alpha = 0.2
160
Alpha = 0.1
140
Sales and forecasts
120
100
80
60
40
20
0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 91011
History in periods
Figure 12.4 Exponential smoothing.
unstable because it would depend too heavily upon demand in the last period.
Conversely, below α of 0.1 the results change too slowly.
Figure 12.4 illustrates three situations. In Figure 12.4a, the demand is quite
stable so all αs give a reasonable forecast, although 0.1 is the most consistent
(below 0.1 it is too insensitive). As the fluctuations are small a larger value of α
could be used in case the demand rate takes off; α = 0.1 is also good where demand
rate is very variable, since the forecast does not fluctuate too much from period to
period (Figure 12.4b). However, where the demand rate changes significantly,
more rapid response to change is needed by including more of the recent demand,
so high values of α are used (up to 0.4), as shown in Figure 12.4c. However, as
shown, the forecast responds to history so it always lags behind the trend.
If little is known about the item, or a single value of α is required for a range of
items, then a median value of α = 0.2 can be used. The value of α should be
reviewed at regular intervals and altered where necessary.
It is quite easy to automate the choice of smoothing constant and thereby
allow each stock line to have an individually chosen α. The technique of ‘focus
forecasting’ will be discussed later.
Exponential forecasting is so easy to put into a spreadsheet.

12.C.1.1 Alternative formula
Another convenient method of updating exponential forecasts uses the error on
the forecast instead of the actual demand. The formula can be re-arranged as
follows:
If new forecast = α × actual demand + (1 – α) × old forecast

then new forecast = α × actual demand − α × old forecast + old forecast
so new forecast = old forecast + α (actual demand − old forecast)
which is new forecast = old forecast + α × (error in old forecast)
This formula is used as frequently as the original one, but it is important when
using it to ensure that any negative signs are included.
In exponential smoothing (and moving average), the forecast for subsequent
periods is the same as that for the next period. It will not predict a trend in demand,
so the accuracy of forecasts further ahead will often be poorer.
12.C.2 Contributions of history

With exponential smoothing, history never gets discarded, it just becomes less
significant. The forecast for this week with α = 0.3 is 30% of last week’s demand
and 70% of history. But that 70% was calculated from 30% of the demand of the
week before and 70% of the previous history, i.e. last week is 30% of the forecast,
1 2 3 4 5 6 7 8 9 10 11 12 13
1
Demand 30% of result
0.9 Forecast 70% of result
0.8
0.7
Fraction of forecast
0.6
0.5
0.4
0.3
0.2
0.1
0
Recent Old
History in periods
Figure 12.5 Contributions of historical data to the forecast.

the week before 21% (30% of the 70%) and previous forecasts are 49% (70% of 70%).
Taking it back further, the contribution from three weeks ago is 14.7% and four
weeks ago is 10.29% (Figure 12.5). In contrast, moving average takes history at
full value until it becomes too old and falls off the end of the calculation.
The contributions of historical demand for different values of smoothing factor
are shown in Figure 12.2. The forecast using α = 0.4 takes little heed of demands
which are more than six periods old (only 4.7%), whereas the forecast using α = 0.1
still has significant contributions (28.2%) from data over 12 months old, and demands
from as long ago as 18 months are significant (15% contribution to the forecast).
12.C.3 Initialising the forecast

The fact that exponential smoothing depends on old history gives a problem when
initialising the forecast. What should be taken as the first ‘old forecast’ in the
equation? There are two ways of solving this: estimate and history.
For a new product type or a new addition to a stock range, it is best to make an
estimate as a forecast. It is not a good practice to start with zero as the ‘old forecast’
since it takes too long to settle down, especially with low values of α. Estimating
can be done by:
• inspired market assessment;

• a model or replication of a similar item;
• an arithmetic average of such data as is available;
• historical forecast from initial demand.
The influence of the estimate will become reduced at the rate shown in
Figure 12.2 or 12.5 depending on the α factor chosen.
Where exponential smoothing is being introduced to an existing product, historical
forecasting can be used. The process is then to make an initial average of the oldest
information (three or six periods simple average) and use that at the first ‘old forecast’.
Then apply this to the oldest demand data, and use the exponential averages succes-
sively for each period until the forecast from the current period is reached (this is
like taking a running jump). This process is illustrated in Figures 12.3 and 12.4. Once
the initial forecast has been established, the exponential average is self-perpetuating.
12.D Improved values for mean absolute deviation

The reason for exponentially smoothing the forecast was to respond to the up-to-date
demand whilst retaining the balance from older data. The same logic should be applied
to the measurement of forecast errors and any other time series used for forecasting.
As customer order patterns change, so should the safety stock. Therefore, an exponen-
tially weighted MAD is required. Updating of the MAD can be done most conven-
iently using the second version of the exponential smoothing formula, therefore:
New MAD = old MAD + β × ε

Where ε is the error in MAD:
ε = Actual Absolute Deivation – Old MAD
and β is the MAD smoothing constant and is the same idea as α, but using β shows
that it is the MAD that is being smoothed rather than the forecast. Note that ε can
have positive or negative values.
It is normal to take β = 0.15 when smoothing the MAD. The reason for this is
that the MAD reflects the characteristics of the demand pattern, and this would
only be expected to alter gradually. If there is a sudden change in the variability of
demand, it is likely to be detected first through a change in demand level and fore-
cast bias. Where standard deviations are used, the weighting has to be applied to
the variance (standard deviation × standard deviation).
Examples
1 MAD forecast = 4, actual absolute deviation = 6
new MAD = 4 + 0.15 × (6−4) = 4.3
2 MAD forecast was = 4 (old MAD)
demand forecast = 17
actual demand for period = 15
so actual absolute deviation = 2
new MAD = 4 + 0.15 (2 − 4) = 3.7
This calculation would be the same if the actual demand were 19.
Although MAD always has a positive value, the error in the MAD can be
positive or negative.
Figure 12.6 shows the construction of the exponentially weighted MAD for the
variable demand in Figure 12.3. This MAD is initialised in the same way as the
exponential forecast from an average applied some periods back (11 periods in
Figure 12.6) and then using the formula successively to get the up-to-date value
(61.7) – another running jump. The most recent result should then be used each
period end to recalculate the safety stock each time. In Figure 12.6 the MAD value
does not change significantly, but in practice demand patterns alter and it is
important to flex the safety stocks using the up-to-date MAD.
12.E Choosing the best forecast – focus forecasting

The question is ‘how to choose the best forecast?’ since the techniques all use the
historical data and give different results. In practice, the system can be used to
choose which forecasting technique or which value of α to use.
Week Demand Forecast alpha = 0.2 Absolute error α error in MAD MAD
Initial 113 60
1 57 102 56 –4 59.4
2 35 88 67 7 60.5
3 171 105 83 22 63.8
4 49 94 56 –8 62.6
5 62 87 32 –31 58.0
6 189 108 102 44 64.5
7 109 108 1 –63 55.0
8 36 94 72 17 57.6
9 30 81 64 6 58.5
10 131 91 50 –8 57.2
11 178 108 87 30 61.7
Beta = 0.15
Figure 12.6 Exponentially weighted MAD.
There are some sensible assumptions to be made:
1 The forecast which was most successful last period will be the best one now.
2 ‘Best’ is defined by historically having the smallest errors.
3 Errors can be calculated as the mean absolute deviation (especially exponen-
tial MAD).
So, the easy way of determining which forecast to choose is to look at the MADs
for each of them and take the smallest MAD as the best forecast. By calculating
all the forecasting options each period, the best forecast for each item can be
made. Items will change forecasting models accordingly as conditions alter.
Exponentially weighted models have the advantage because they are more sen-
sitive to recent events whilst still influenced by older data. They are generally
preferred over the simple averaging techniques such as moving average, regres-
sion or trend measurement.
Produce many forecasts and choose the one with the smallest MAD.
Choosing the best forecast is therefore usually carried out using exponentially
weighted MADs, as a more reliable method of assessing the potential of alterna-
tive forecasts. The chosen forecast is used until the end of the period, then the
comparison is repeated.
Period Demand Alternative forecasts

1 28 Model type Parameters Forecast demand MAD
2 45
3 33 Moving 12 month 26.8 5.9
average
4 39 6 month 22.0 7.4
5 17 α = 0.1 28.3 9.5
6 27 Exponential α = 0.2 24.7 6.2
smoothing
7 22 α = 0.3 23.8 5.7
8 24 α = 0.4 23.9 5.2
9 15 Double α = 0.2 21.9 5.4
exponential
10 18 smoothing α = 0.3 21.8 5.0
11 28 α = 0.4 21.4 5.7
12 25
Figure 12.7 Focus forecasting.

Note: α = 0.1 for DES gives very little variation and is therefore not included.
Double exponential smoothing calculation is discussed in Chapter 13.
Example
The demand pattern for an item over the past 12 periods is shown in the
two left hand columns of Figure 12.7. Several alternative forecasts
were calculated from this history, as shown together with the MADs.
The smallest MAD is 5.0 and this is taken to be the most accurate fore-
cast, which is double exponential smoothing with α = 0.3, and the worst
forecast is exponential smoothing with α = 0.1 (MAD = 9.5). Therefore,
the forecast to choose is 21.8. However, at the next period end a differ-
ent forecast model might well prove to be the most accurate.
12.E.1 Forecast tracking using cumulative sum of errors

Historical forecasting reacts to demand and is therefore always lagging behind.
If there is a gradual increase in demand level, the forecast will be consistently too
low (or for a decrease the forecast will be high).
The forecasting methods discussed will cope with most demands, but they need
changing from time to time when the method is inappropriate to the conditions.
Therefore, a monitoring process is needed to enable corrective action to be taken.
Tracking signals provide a monitor of how well the forecast is performing.
When the errors fall outside acceptable limits, the tracking signal identifies the
problem or modifies the controls.
The Cusum technique provides a warning signal when there is a significant bias
in the forecast (i.e. it is always too low or too high). It is very simply obtained by
accumulating the errors in forecast over successive periods. When the forecast is
correct, the actual demand is distributed equally above and below the forecast and
cumulative error is minimal. Where the forecast is consistently too low or high,
the total error (Cusum) gradually builds up.
If this Cusum becomes unacceptably large, an error signal can be triggered and
the forecast examined. What is unacceptable? If the forecast is always too low,
then the calculation of Cusum gives the same result as the calculation for MAD.
Consider the formula
Cusum
u
Tracking Signal =
MAD
Ideally, for an accurate forecast, this ratio is zero. In the worst case, the results
are the same size and the tracking signal is +1 or –1 (depending whether the fore-
cast is too high or too low). As a working assumption, any forecast with a tracking
signal of more than 0.4 should be scrapped.
Get the system to identify bad forecasts.
Tracking signals are illustrated in Figure 12.8 for variable and rising demands
(shown in Figure 12.4). For the variable demand, the tracking signal does not indi-
cate a problem, but when rising demand is established, the tracking signal becomes
too large (> 0.4). Here, for better results, exponentially weighted averages are used.
The smoothed cumulative error, S, is calculated in the same way as exponential
MAD, without using the absolute error value
St = δ × et + (1 – δ) × St–1
where S is the standard error, e is the actual error and δ is the smoothing factor for
the errors. The subscripts on the standard errors, t and t–1, stand for this period
just completed and last period. δ is set at the same value as β, 0.15 (or 0.1).
Remember:
MADt = βet–1 + (1 – β) MADt–1.
When a problem signalled by Cusum has been resolved, the Cusum should be
reset to zero, otherwise the error signal will remain high.
Tracking signals are very useful and allow system generated forecasts to monitor
themselves, self-correct and warn inventory managers when it is producing bad
forecasts. The tracking signal can be used simply as a warning flag and create an
exception file and a manual adjustment, or, better, the system can select an alterna-
tive forecast.
12.E.2 Achieving the best results

This combines the tracking signal with focus forecasting, the optimum forecast
being selected by choosing the one with the smallest smoothed MAD as normal.
If this forecast has a tracking signal below, say, 0.4, then the forecast is acceptable.
If the tracking signal is greater than the unreliability level (0.4), then the forecast
with the next smallest MAD can be chosen. This is a straightforward, mechanical
and effective way of getting good forecasts across the range of stock items.
12.E.3 Trigg’s tracking signal

When Cusum indicates a significant forecast error, the forecast needs changing
rapidly. One approach for exponentially weighted averages, is to use a larger
smoothing constant, α (see section 12.C.1). Where focus forecasting is not available,
Trigg’s method of feeding back the Cusum information automatically is useful.
This technique uses the tracking signal generated by exponential smoothed Cusum
and MAD (as in Figure 12.8). The tracking signal is then used as the α factor. This
is an elegant approach.
Trigg’s method avoids multiple calculations.
Accurate forecasts give low values for tracking signals, and need little modifi-
cation, so low values of α are best. Large tracking signals indicate significant
forecast errors, so high α factors are required to alter the forecasts. This suggests
that tracking signals can be used as smoothing constants for the forecasting, and
enable exponentially weighted forecasts to be more self-compensating each per-
iod. For this purpose, the tracking signals have to be positive numbers and so any
negative tracking signals (TS) are changed to positive. The forecast, F, is then
given by the usual formula:
Ft + 1 = Ft + |TSt| et
where et is the forecast error for the latest period, and the value of TS is always
positive. TS can be used for feedback into exponential smoothing or double expo-
nential smoothing. It is also possible, but less beneficial, to calculate a tracking
signal using normal averages. The values of δ and β have to be the same, which is
normally 0.15 for weekly demand and 0.1 for longer periods (e.g. monthly) or
there is a danger when using Trigg’s method that the smoothing factor becomes
very high (α > 0.4) when it tends to overcompensate for small changes. Previous
Variable demand Rising demand
Period Demand Exp. Tracking Exp. Tracking Demand Exp. Tracking Exp. Tracking
error signal error signal Error signal Error signal
α = 0.2 α = 0.4 α = 0.1 α = 0.3

Initial 0.00 0.0 0.00 0.0 47 0.00 0.0 0.00 0.0
1 57 –9.96 –0.2 –10.97 –0.2 72 –0.81 0.0 –0.81 0.0
2 35 –20.15 –0.4 –18.92 –0.3 67 2.33 0.1 1.97 0.1
3 171 –6.74 –0.1 –3.12 –0.1 82 3.95 0.2 2.21 0.1
4 49 –11.63 –0.2 –6.31 –0.1 104 7.38 0.3 4.73 0.2
5 62 –16.32 –0.3 –15.39 –0.3 31 13.19 0.5 8.79 0.4
6 189 0.42 0.0 1.88 0.0 90 6.49 0.2 –0.06 0.0
7 109 3.60 0.1 5.99 0.1 91 10.12 0.4 5.37 0.2
8 36 –7.70 –0.1 –7.52 –0.1 105 12.89 0.5 6.77 0.3
9 30 –18.24 –0.3 –16.88 –0.3 92 16.92 0.6 9.35 0.4
10 131 –9.89 –0.2 –6.08 –0.1 113 17.80 0.6 8.47 0.4
11 178 6.17 0.1 10.49 0.2 100 21.35 0.7 11.09 0.5
Unacceptable forecasts are shaded.
Figure 12.8 Tracking signals.

45
40
Demand and forecast values
35
Demand
30
Exponential
25 Forecast α = 0.2
20 Trigg's
Exponential Forecast
15
10
5
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Figure 12.9 Trigg’s method.
discussion stated that α should be less than 0.4 for exponential smoothing,
although higher values are acceptable for double exponential (because α is
squared). It is therefore advisable in practice to modify the formula so that large
values of the smoothing factor are not created.
Example
Figure 12.9 shows the exponential forecasts with α = 0.2 and using
Trigg’s method. The characteristics of the demand pattern change after
period 8.
The exponential forecast gives a reasonable forecast up to period 7;
the data is shown in Figure 12.10a. The smoothed error, which was very
small up to period 9, starts to increase. This is because the change in
forecast lags behind the actual demand level. Consequently, the tracking
signal increases. It reaches 0.4 by period 10, which is quite quick. At
that stage, the forecast can be readjusted or an alternative model used,
otherwise poor results will continue.
In Figure 12.10b the tracking signal is fed back to change the α as
proposed by Trigg. The α factors are much smaller at the beginning. For
the first few periods this is because initially the smoothed error was
assumed to be zero. The tracking signal stabilises and responds less to
the low demand in period 7, which is sensible given the consistently
larger demand in the previous periods. When the demand level changes,
12.10a Exponential forecasting 12.10b Trigg’s method
Period Demand Normal Error in Smoothed Smoothed Tracking Trigg’s Error in Smoothed Smoothed Tracking
exponential Forecast MAD error signal exponential Forecast MAD error signal
forecast forecast
α = 0.2 β = 0.1 δ = 0.1 β = 0.1 δ = 0.1
1 13 16 −3 4 0 0 16 −3 4 0 0
2 1 15.4 −14.4 5.04 −1.44 −0.29 16 −15 5.1 −1.5 −0.29
3 10 12.52 −2.52 4.79 −1.55 −0.32 11.59 −1.59 4.75 −1.51 −0.32
4 17 12.02 4.98 4.81 −0.89 −0.19 11.08 5.92 4.87 −0.77 −0.16
5 11 13.01 −2.01 4.53 −1.01 −0.22 12.02 −1.02 4.48 −0.79 −0.18
6 19 12.61 6.39 4.71 −0.27 −0.06 11.84 7.16 4.75 0 0
7 16 13.89 2.11 4.45 −0.03 −0.01 11.84 4.16 4.69 0.42 0.09
8 15 14.31 0.69 4.08 0.04 0.01 12.21 2.79 4.5 0.66 0.15
9 26 14.45 11.55 4.82 1.19 0.25 12.62 13.38 5.39 1.93 0.36
10 38 16.76 21.24 6.47 3.2 0.49 17.41 20.59 6.91 3.79 0.55
11 31 21.01 9.99 6.82 3.88 0.57 28.72 2.28 6.44 3.64 0.57
12 40 23.01 16.99 7.84 5.19 0.66 30.01 9.99 6.8 4.28 0.63
13 35 26.4 8.6 7.91 5.53 0.7 36.3 −1.3 6.25 3.72 0.6
14 32 28.12 3.88 7.51 5.36 0.71 35.52 −3.52 5.98 3 0.5
15 28.9 33.76
Figure 12.10 Forecasting with Trigg’s method.

the tracking signal increases and enables the forecast to rise faster to
cope with the problem. However this does not fully compensate and the
tracking signal still rises above 0.4. The forecast rose quickly to the new
level as the α factor increased (see Figure 12.10). It did this without the
inventory controller having to intervene, as can be seen in Figure 12.9.
With Trigg’s tracking signal, the alpha factor can go above the 0.4 level
since it can then settle down to a reasonable level when it has adjusted
(although there is some danger of it oscillating as a consequence). The
tracking signal is still rather high in period 19. As the smoothed error
continues to decrease because the forecast is again accurate, this
tracking signal will reduce gradually.
The data represents a typical demand level change. Trigg’s approach is not the
ideal answer, since it lags behind the demand pattern significantly in a period of
change, but it is an improvement where there are many stock items and little time
to manage them individually. However, if the alpha factor stays high for many per-
iods, then manual readjustment is necessary.
Key points
• Historical forecasting is essential for basic stock control.
• It is necessary to add information about future events.
• A rolling average gives a basic forecast.
• This can be greatly improved using exponential smoothing.
• Carry out several forecasts, then select that with the smallest MAD.
• Tracking signals shows when forecasts are biased.
13 Improved forecasting methods
• Better forecasting for rising or falling demand.

• Coping with seasonal demand.
• Alternative techniques.
13.A More forecasting tools
13.A.1 Developing exponential forecasting

Exponential smoothing is the basic model, which can, of course, be improved.
It splits demand into two components:
• Average, which is re-calculated each period.

• Random fluctuations about the averages, discussed in Chapter 7.
The next stage of improvement of forecasting can be to look at simple

approaches to changing and cyclic demand patterns. Two common situations that
can be addressed are:
1 a constant increase or decrease in demand, using double exponential

smoothing, or
2 a seasonal variation of demand, using base series forecasting.
These then form the basis for more complex models. In general, the more complex
the model the more history is required. The constraints on forecasting are often
practical ones:
• insufficient history and short product life cycle;

• changing demand patterns;
• one-off events;
• change in competition;
Improved forecasting methods 189
• external fashion, economic and regulatory influences;
• inaccurate data;
• misleading market information.
These are usually the limiting factors on forecast accuracy. Attention to these
issues are often more effective than employing highly sophisticated mathematics.
It is a balance between time, cost and understanding. If the system includes a
forecasting tool, then it should be tested and used. Understanding how to optimise
it is important, since forecasts are tools for the user. In many situations, the simple
use of seasonal and trend forecast give good results.
13.A.2 Double exponential smoothing

When supply lead times are long, it is particularly important to detect any trends
in demand, otherwise excess inventory or shortages will occur. It is easy to con-
sider a gradual consistent linear increase or decrease in demand. The techniques
in Chapter 12 establish a demand level, assuming that demand rate will stay at the
same level as the latest forecast. The discussions show that this is a dubious
assumption, and therefore updating the parameters regularly is essential.
Double exponential smoothing adds a trend factor to single exponential
smoothing so that demand is analysis into ‘average + trend + fluctuations’. This is
a simple uniform increase or decrease; Figure 13.1 shows the difference.
110
Demand
alpha = 0.4
100
alpha = 0.2 Forecast with trend
90
80
70
Forecast average
60
50
40
30
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Figure 13.1 Improved forecasting through double exponential smoothing.

To understand double exponential smoothing, it is best to go back to simple
mathematics. Figure 13.2 shows a normal straight-line graph given by the formula
y = bx + d with slope b and cutting the y axis at point D. The double exponential
model is equivalent to this straight line, with y being the demand and x being time.
Single exponential smoothing is like the line at D above the x axis.
In double exponential smoothing the forecast of demand, Ft , is therefore com-
posed of two factors, Ft +1 = dt+ bt . The demand level is dt, similar to single expo-
nential smoothing. The second factor calculates the trend of variations and adds a
‘slope’ factor. The actual demand will also include a factor caused by random
demand fluctuations.
Double exponential smoothing (DES) has important advantages because the
extrapolation to future periods continues, the increase or decrease is quantified
rather than assuming a static demand level. In addition to this, DES can also com-
pensate for lag in the forecast. It is obvious in Figure 13.1 that the forecasts are
lagging behind an increasing demand level (and will signal a tracking error). DES
can bring the forecast in line with the demand once a linear trend has been estab-
lished. This can be very useful when there is a long-term increasing or decreasing
average demand.
The lag in an exponentially weighted average is simply given by the expression
1− α
Lag Trend
α
where α is the smoothing factor.

From Figure 13.2 it can be seen that the double exponential forecast for one
time period ahead is d + b (demand + trend). The forecast for two periods ahead
will be d + 2b (demand + twice trend).
There are several versions of the double exponential formula. The simpler
version is Brown’s model. A more comprehensive model is Holt’s model, but this
Ft
y
or
Demand
y = bx + d
b
D
1 2
x or Time periods
Figure 13.2 Double exponential model.

1 Find the forecasting error

error = forecast – actual demand
2 Update the demand level
demand level = actual demand + (1 – α)2 error
3 Update the trend
new trend = old trend – α2 error
4 Add the components
new forecast = demand level + new trend
To update the forecast requires values for:

• existing trend
• previous forecast
• actual demand.
Figure 13.3 Double exponential smoothing process.
requires two smoothing constants (one for the base forecast ‘d’, and one for the
trend ‘b’). Holt’s method does allow optimisation of the smoothing factors for
selected items by iterative programming.
The simpler formula is adequate normally and uses only one smoothing factor,
which can be denoted as α since it behaves like the single exponential smoothing
factor. The method of calculating values of ‘d’ and ‘b’ and the forecast is shown in
Figure 13.3. With Holt’s method, different values of α are used for demand level
and trend.
The general forecast for n periods ahead is:
Ft + n= dt + (bt × n)
Using this linear extrapolation of trend, a much more accurate forecast of demand
in future periods can be made, if a gradual increasing or decreasing sales pattern
is expected (see Figure 13.4).
For items with very variable demand patterns, the estimation of trend can be
misleading. An extreme demand value for the previous period reflects too heavily
on the calculated trend. If α is reduced to compensate, then the model becomes
unresponsive. In this case it is better to take a longer sample period to even out the
statistical variations, and this is often done by calculating the average demand over
the last three periods and using the averaged error to modify ‘d’ and ‘b’. The trend
calculation is smoothed through the demands over preceding periods. Therefore,
the greater the smoothing, the slower is the trend to respond to change in trend.
Changing forecast methods to DES means that the initial values of ‘d’ and bt–1
have to be estimated. The smoothing constant behaves like the smoothing constant
for single exponential smoothing. The larger the value, the faster it responds to
change, but the more variable is the result.
Week Customer Error Demand level Trend Forecast for
demand recalculated, d recalculated, t next week, Dt–1
9 61.34 2.57 63.91
10 91 −27.1 73.66 3.65 77.31
11 40 37.3 63.88 2.16 66.04
12 56 10.0 62.42 1.76 64.18
13 56 8.2 61.23 1.43 62.66
14 35 27.7 52.70 0.32 53.02
15 80 −27.0 62.74 1.40 64.14
16 85 −20.9 71.65 2.23 73.88
17 43 30.9 62.76 1.00 63.76
Figure 13.4 Double exponential calculation.
140
120
100 Forecasts
Demand and forecast values
80
60
40 Demand
alpha = 0.4
20 alpha = 0.2
Exponential
smoothing alpha = 0.4
0
1 2 3 4 5 6 7 8 9 10 11 12 13
Period
Figure 13.5 Double exponential smoothing forecasts.

As the smoothing constant is squared for DES, selection of the correct
value is more important (this can be done automatically by focus forecasting).
In Figure 13.5 some historical and future results demand are shown. DES with
α = 0.2 appears the most reliable and forecasts an increasing demand of 3.2 items
per period. Single exponential smoothing just estimates a static demand at 92,
which would be reasonable for many situations where the demand level is not
changing, unlike the pattern in Figure 13.5.
Starting exponential smoothing forecasts requires estimates of ‘d’ and ‘b’ and
this depends upon the information available. If the demand for one period only is
known, or there is an estimate, then there is little alternative to setting ‘d’ equal to
that value, ‘b’ to zero or an estimate and using a high value of smoothing constant
so that the forecast will respond rapidly to the actual trend. Thereafter, the smooth-
ing constant can be reduced to a more stable value.
For an established product, the history is known and the values of ‘d’ and ‘b’
can be estimated accurately. In this case it is best to plot a graph and estimate the
value of demand, Ft, and a trend ‘bt’. Then
‘dt’ = Ft – bt
These values of ‘d’ and ‘b’ can then be used to create double exponential forecasting.
Use double exponential smoothing where MAD is small compared to demand.
60
alpha = 0.3
Demand 3
50
40
2
30 Forecasts
20 1
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Figure 13.6 Effects of sudden demand changes.

It can be seen that DES has significantly beneficial effects. However, it is like
carrying a long ladder: a small movement causes the extremities to swing a
large amount! Where there are sudden changes in demand rate (Figure 13.6),
the forecast needs interpretation, especially where there are long lead times and
there is:
1 rapid change in trend ‘b’;

2 a move to another level of demand (e.g. major sales account gain or loss).
1 Rapidly changing trends: The forecast will not respond immediately to sudden
increases or decreases in demand rate, the next forecast will underestimate
the change (Figure 13.6 situation c), and further forecasts exaggerate the
trend, ‘b’, in order to give a good forecast for subsequent periods. This over-
estimation of trend is not sustained, but long-term forecasting using it until it
settles down would be unwise (Figure 13.6 situation d). A practical solution
is to use the calculated ‘b’ value for the short term, but to use a more histori-
cally smoothed trend value for the long term.
2 Shift in demand level: Rapidly increasing demand cannot continue for ever, so
the linear trend will cease. The DES forecast will take time to adjust and
excess demand will be forecast. A time limit is therefore normally set and the
trend reverts to a level demand at this horizon limit (at e in Figure 13.6). This
horizon depends on the characteristics of the item lifetime, but for many
industries is half a year.
There are alternative formulae for DES available on media which give equally
valid forecasts. The capability to optimise the smoothing factors for demand level
and trend gives significant improvements in forecasts.
13.B Forecasting for seasonal sales

Many products have seasonal or cyclic demand patterns. These can be caused by
natural seasonal factors – for example, temperature, rainfall or the probability of
thunder storms – or by other seasonal factors – such as holiday patterns, Christmas
or other religious festivals, or financial year-ends. These factors can be the
overriding cause of demands, or small contributions for fine-tuning the forecast.
Seasonal demands can be masked by the sporadic nature of demands for an item,
as is often the case for C class items. It may well be expedient to ignore the
seasonal variation and treat this as a random fluctuation too. Here a 12-month
moving average is a simple solution, although the safety stock will, of course,
inflate the service outside the faster-moving season and cause more stock-outs
during the season.
A better solution is to take the monthly sales figures for the whole sector or
company. This will not suit all the products, but is good for the majority and it
gives a smooth reliable result, although not optimised for individual items.
Use specific forecasting models for seasonal demand.
The general approach to seasonal demand is to determine the seasonal pattern

and then nullify it. This then leaves the forecaster free to use any other technique.
The result can then be seasonally adjusted again.
13.B.1 Historical data

Forecasting seasonal demand requires at least a year’s worth of history or an
estimate. In fact, the more years for which data are available, the more reliable the
forecast (unless there has been a change in the pattern of requirements for the
item). This is usually a matter of interpretation, since the data is often distorted by
specific, non-repeating factors to do with the market, large contracts or price
changes at irregular intervals. The historical data often has to be interpreted to get
the best forecast.
Generally, seasonal data is collected in monthly buckets. For most purposes,
forecasting can be done on a weekly basis, but the demand smoothing by using
months is advantageous in calculating reliable seasonal factors. However, where
Christmas dominates the sales, weekly information around November and
December is often needed.
Ideally, the demand for each corresponding month for three or more years is
averaged so that an average monthly demand pattern emerges (a weighted average
biases the data in favour of the most recent years and gives better results). With less
historical data, the reliability of the results is reduced.
13.B.2 Base series technique

The classic technique for calculating seasonal demand is base series forecasting
(an alternative technique is ‘regression analysis with dummy variables’, which is
similar but does not benefit from the exponential weighting). Base series produces
seasonal demand forecasts as an extension to exponential smoothing. It makes the
assumption that the seasonal variation calculated from history can be applied
accurately to the current year, and that forecast errors are due to changes in the
average demand level or to random fluctuations. The process for updating the
forecast is:
1 Take the new month’s (or week’s) demand.

2 De-seasonalise it.
3 Compare it with the forecast.
4 Update the forecast for the next months (exponentially).
5 Re-seasonalise the forecast.
Demand history Seasonal factors Base

series
Month Year 1 Year 2 Year 3 Year 1 Year 2 Year 3
1 40 60 60 0.421 0.600 0.571 0.531

2 20 20 40 0.211 0.200 0.381 0.264
3 80 70 60 0.842 0.700 0.571 0.705
4 200 150 180 2.105 1.500 1.714 1.773
5 150 230 240 1.579 2.300 2.286 2.055
6 210 240 270 2.211 2.400 2.571 2.394
7 140 160 200 1.474 1.600 1.905 1.659
8 70 20 60 0.737 0.200 0.571 0.503
9 30 30 0 0.316 0.300 0.000 0.205
10 10 20 10 0.105 0.200 0.095 0.134
11 70 100 50 0.737 1.000 0.476 0.738
12 120 100 90 1.263 1.000 0.857 1.040
Total 95 100 105 1 1 1
Figure 13.7 Creating the base series.
Before the forecasting can start, seasonal demand factors have to be established
and an initial forecast made for the average demand. Creating the seasonal pattern
in shown in Figure 13.7 (there is a lot of data in this figure, but the process is sim-
ple). Dividing the actual demand by the average monthly demand (AMD) gives
the relative size of the monthly demand, and the averaging each month gives the
base series (in practice, the AMD allows for trend in underlying demand rate and
the averaging is done exponentially).
Having created the base series, and the current month’s forecast, the forecasting
processes 1 to 5 are then carried out. In Figure 13.8 the initial month’s forecast is
130. Since the base series for month 1 is 0.53, the expected sales would be 69.
However, at the end of the month it was found that the actual sales were only 50,
corresponding to sales in an average month of 94.2 (= 50 / 0.531). The expected
sales were 130, so the forecast was too optimistic. Any forecasting technique
can now be used as the forecast is not seasonal. In this example, exponential with
α = 0.2 is chosen. The result is therefore 122.87. This is the new forecast to use
for month 2 where the base series is 0.26, which means an expected sale of 32.4.
By the month end, the sales were 30, equivalent to 115.4 in an average month.
Exponentially smoothing the demand of 115.4 against the forecast of 122.87 gives
121.37 for the new forecast. As in normal exponential smoothing, this process is
carried out at each period end.
Seasonal Base series Equivalent

Forecast Actual New smoothed average
month from average
sales sales Figure 13.7 demand
Initial Forecast = 130.00

1 68.90 50 0.53 94.3 122.87
2 31.95 30 0.26 115.4 121.37
3 84.96 70 0.7 100.0 117.10
4 207.26 260 1.77 146.9 123.06
5 252.27 220 2.05 107.3 119.91
6 286.58 2.39
7 1.66
8 0.5 Normal exponential
9 0.21 Smoothing alpha = 0.2
10 0.13
11 0.74
12 1.04
Calculation
Base series X Exponential calculation
Actual sales From historical Actual sales divided
new smoothed from forecast and
recorded base series by base series value
average equivalent actual
Figure 13.8 Forecasting with base series.
Example
Demand for July is forecast as 36. This is because the average monthly
demand (AMD) is 30 and history has shown that the demand next month
is typically 20% above average. At the end of the month, the records
showed that 33 were sold. What is the likely demand for August, which
has typically proved to be 10% above average?
The demand in July should have been 30 x 1.20 = 36, but instead
it was 33, which is 27.5 x 1.2 (the assumption is that the seasonal
factor is correct). The de-seasonalised forecast should have been
30, but was 27.5. Therefore, the forecast was too high and can
be exponentially smoothed downwards (assuming α is 0.2 here) as
follows:
New de-seasonalised forecast = α × de-seasonalised demand
+ (1 − α) × old de-seasonalised
forecast
= 0.2 × 27.5 + (1−0.2) × 30
= 29.5
If the de-seasonalised forecast is 29.5, and the demand for August is

10% above average, then the new expected sales for August are 29.5 +
10% of 29.5, i.e. 32.45.
The de-seasonalised demand forecast of 29.5 can be used to forecast
all future months by multiplying by their respective seasonal factors.
13.B.3 Other forecasting methods

Exponential forecasting techniques have proved to be successful in ensuring stock
predictions are good in practical situations which generally suffer from poor data
integrity, feedback from the market and changing demand patterns. Although
incorrect data is a prime problem, better forecasting methods work better with
clean data, including the
• measurement of demand, not sales;

• segregation of demand for an item into ex-stock, routine schedules and large
project requirements (usually buy against customer order);
• time bucket control, variations in the working week caused by holidays, etc.
Market input into forecasting is often problematic, but it is essential for ‘events’.
Using the Delphi Method is reasonable (i.e. get several experts to study the market
individually, then get them together and come to a consensus). However, it is still
likely to be qualitative and more suited to long-term generic forecasting.
There are several alternative qualitative techniques that are employed,
including:
• Unweighted average trends (regression analysis): It is important that outliers

are eliminated before using this and other techniques. This technique is not
favoured because it is not weighted.
• Probability-based forecasting (Baysian forecasting): Pure mathematics
turned practical technique. The illustration in Figure 13.9 shows four options
for the reason for increased demands:
changes in levels;
once-off spikes;
trends;
random variability.
With the information available, these are equally likely (25% chance of
each). If the demand stays at this inflated level at the next period end, then
‘change of level’ is now more likely, but the others are less likely, espe-
cially random. New data alters the probabilities and therefore composes a
new forecast.
Figure 13.9 Baysian approach.
Forecasting model Application Constraints

Moving average Static demand level, irregular Usually unsuitable for
pattern major products
Regression analysis Continuous trend, irregular Oversensitive to outliers
pattern
Exponential smoothing Variable demand, gradual Good basic workhorse
changing level, and slow
moving spares items
Base series Seasonal demand, mature Basis for most seasonal
products with low variability forecasting
Double exponential Consistent increasing or Apply where variability
smoothing decreasing demand, does not mask trend
product phase-in and
obsolescence
Fourier analysis Cyclic established demand Product with long consistent
e.g. commodities history
Baysian forecasting General products Mathematics needs
developing
Fuzzy logic Sophisticated modelling Not yet available
More sophisticated Improved exponential Much history often needed,
techniques forecasts with more for longer term forecasting
complex formulae
Specific models Based on market knowledge, Unreliable due to
potentially most accurate forecaster’s bias
Figure 13.10 Forecasting models.

• Curve fitting (Fourier analysis): Musical instruments sound different because
of harmonics – sine waves. Demand patterns can similarly be analysed into
sine waves and then the pattern extended to give a forecast. Ideally, this
forecasts all demand, including seasonal variations. As it requires a lot of
history to establish the pattern, the applications are somewhat scarce.
• Extended exponential methods: Obviously, the linear trend assumption can be
improved by making it into a curve. The mathematics become increasingly
complicated in the work originally by Wagner, Within, Holt, Winters and
others. Practitioners should grasp models including these methods if they are
optimised and used in competition with simpler options.
• Other options have been developed for specific situations, such as Croston’s
method for slow-moving items. With the development of co-operation and
reduced lead times along the supply chain, there is the possibility that the
demand for some companies could change from independent to dependent,
avoiding the need for forecasting. Accessing ‘Big Data’ and structuring the
supply chain through demand driven MRP can eliminate the need to forecast
and hold inventory other than critical points. In the future, mathematicians
may be able to apply fuzzy logic to forecasting and stock and improve optimi-
sation against the KPIs.
Figure 13.10 shows forecasting techniques and their merits.
Key points
• Double exponential smoothing gives tangible benefits where there
are:
o significant trends in demand;

o long lead times for A class.
• Double exponential works best where demand is consistent.

• Seasonality has to be taken into account in the forecast.
• Base series is a simple effective technique which links to other fore-
casting methods.
• Use and manage more sophisticated techniques that are provided
in software.
• Use focus forecasting continuously to assess the best forecast.
More information on these techniques and the latest developments are
on the internet.
14 Dependent demand
• ERP and dependent demand.

• Planning material requirements (MRPII).
• Structuring demand in the supply chain.
• Interpreting bills of materials.
• Master scheduling – the key to success.
• Triggering orders – projected available.
• Meeting demand – available to promise.
14.A Avoiding uncertainty

Many companies have invested in ERP (enterprise resource planning) systems to
run their business. These attempts to provide a comprehensive system for all
aspects of the business are particularly good in the accounting areas for individual
and group companies. The database structure makes them flexible, so that users
can query all aspects of the operations to which they have access.
ERP grew out of MRPII (manufacturing resource planning) and it is this aspect
of ERP that is of major interest because it can be beneficial throughout the supply
chain. The co-ordination of activities to meet customer service is the objective of
MRPII.
Taking a simple view of the supply chain, once the end user has bought an item,
this information should be transmitted to everyone back down the supply chain,
who then replenish what was sold. Why doesn’t this work? The main problems are
delay (lead times) and batching, caused by stockholding and economics. However,
the closer a supply chain gets to this ideal, the better it works. Some places in
supply chains work like this, especially within manufacturing and kanban supply.
Instead of having to forecast (independent) demand, the demand is ‘dependent’
and can be calculated. The contrast between these two approaches is shown in
Figure 14.1.
Dependent demand says ‘get all the items you need together, when you need
them, then sell them, don’t hold stock’, but independent demand can happen at
any unexpected time, so inventory is always required depending on the demand
202 Dependent demand
INDEPENDENT DEPENDENT
Stock level system MRP system
Forecast Calculated
Keeps stock Buys when required
All lines separate Lines co-ordinated
Reactive Proactive
Good customer service Very good customer service
High stock Low stock
Figure 14.1 Contrasting independent and dependent demand.
Dependent demand
Stock supplied when Independent demand
required and used up Delivery Quantity Stock
+ Safety Stock
Average independent
demand inventory
Inventory
Average independent
demand inventory
Time
Figure 14.2 Inventory and MRP.
for each item separately. The more demand is understood to be dependent, the
better the effectiveness ratio,
availability
R=
inventory
r value
For kanbans, R is very high, good for material requirements planning (MRP), but
poor for normal inventories. This is illustrated in Figure 14.2 where the MRP
time interval between incoming delivery and product despatch governs R. This is
the lead time, which in manufacturing can be considered for the whole factory or
individual processes.
Life is simpler if you know what the demand is!
The essential feature of dependent demand is that it is calculated from the

demand of the next item along the supply chain; the aim is to have stock when it
is to be used and no stock the rest of the time. This is the same as ordering for
supply-to-order or make-to-order items where the demand is certain before
supplies are triggered.
Dependent demand 203
14.B Material requirements planning
14.B.1 The structure

The technique for planning dependent demand is MRP. This is a generally
applicable technique for all types of dependent demand. The basic concept is to
have stock when it is needed and to have none the rest of the time. With dependent
demand, the size and timing of the requirements are known from the next level
along the supply chain (the customer). The structure of MRP makes it eminently
useful within manufacturing, especially where there can be several levels of
dependent demand in a product (assemblies made from sub-assemblies made
from components made from purchased materials). Here, MRP is well developed
and complex. In many inventory situations, MRP is a much simpler process.
Manufacturing plan customer orders and so the requirements within production
are defined by this in terms of quantity and time to enable them to achieve it. This
approach can be used for all items, including customer schedules and non-stock
products.
This requires planning to be started with the output (customer demand). Back-
ward scheduling provides a simple process for taking the delivery date required by
the customer and working back to identify when the item, or its constituents, need
ordering from the supplier. ‘Offsetting by the lead time’ is the jargon used for this
delay.
Figure 14.3 shows a sequence of operations where the arrival of the supply from
the previous process is confirmed, then the ‘order’ is auctioned. However, by
synchronising the plans, the supply lead time can be eliminated and the process
lead time (which includes both the processing time and the waiting time in the
process) can be reduced. If the ‘issue’ for Process 3 is at the time to satisfy
customer demand, then the previous operations can be ‘ordered’ at this time and
delays avoided.
If the integrated system is working properly, each process can be initiated just
as the supply arrives (marked ‘receive’ on the diagram). Going back further by the
supply lead time identifies when to place the purchase order.
Control using dependent demand saves money and management time.
The advantages of dependent over independent demand inventory control are

listed in Figure 14.4 and illustrate why inventory managers should seek to identify
dependencies wherever possible.
14.B.2 Planning using MRP

ERP co-ordinates the inventory systems with the sales and financial plans of the
company. Within this framework, MRP is a detailed planning tool based on
Order Receive Issue
Supply Process
lead lead
Order Receive Issue

time time
Transfer
lead
Process 3
time
Supply Process
lead lead
Order Receive Issue

time time
Transfer
lead
Process 2
time
Supply Process
lead lead
time time
Time
Process 1
Figure 14.3 Process sequence.
Independent demand Dependent demand

Inventory-level system Material requirements planning
Treats each part individually Deals with structure
Depends on demand history Looks at future plans
Assumes average Handles erratic demands
Aims to keep stock levels up Holds stock only to cover demand
Priorities inflexible Sensitive to priority changes
Mainly runs itself Needs managing
Figure 14.4 Contrasting dependent and independent demand approaches.
knowing the quantity and timing of the demand, the components of the product
and what is currently available. The formal list (shown in Figure 14.5) is:
1 A master schedule of planned supply to customers for each product, projected

far enough ahead to permit ordering of bought-out items, raw materials
and components. This ‘planning horizon’ must be long enough to cover the
procurement lead time for bought-out items plus the total manufacturing
time.
2 A well-defined ‘bill of materials’ showing purchased items and manufactured
components to identify what to make and buy (usually including packaging
items and anything else which needs to be included in the delivery to the
customer).
3 Lead times for the purchase or manufacture of each component, realistically
assessed. The cumulative supply time will then be understood.
4 Accurate records of stock, including work-in-progress and on-order parts.
5 Operating constraints, including the sequence of processes and the desirable
quantities to be processed at each stage.
The two main attributes of MRP, namely, ‘time phasing’ and ‘structure’, are
now being applied in a wide variety of stocking situations outside manufactur-
ing. If it is known, or can be forecast, when major customers want their supplies,
any business can create a master schedule, with the consequential reduction in
stocks.
Long-term
plan (S&OP)
Demand
forecasts
Product
inventory
Master
schedule
Sales orders
Dependent
Bill of
inventory
materials Material
(WIP, RM)
requirements
plan
Lead times
and routings Batch rules
Supply orders Production

and schedules plans
Figure 14.5 MRPII structure.

Normal stock management treats each item as completely independent,

unaffected by usage rates of any other item and the demand is assumed to happen
at random. The MRP approach is entirely different, based on:
• interdependent usage rates between stock lines;

• acquiring stock for when it will be used.
Many customers tend to use items together, so demand is not truly indepen-
dent. For example, there must be a connection between the demand for hinges
and doors, paint and paint brushes, nuts and bolts, although in general business
there is no certainty that the sale of one item will be matched by the sale of the
linked item. If a customer buys 100 bolts, there will be no automatic order for
100 nuts. They may already have a stock of nuts, or use some in threaded holes,
or even buy from elsewhere. The situation will become clear after a while.
For example, it may work out in the long term that for every 100 bolts bought by
a customer, they require 40 nuts of one type and 20 of another. This relationship
could be consistent, because the customer’s use is consistent, sometimes without
identifying the reason. The demand for these items could therefore be inter-
linked, and variations could mainly result from changes in the overall level of
business. It is important, therefore, that the stock quantities for these should
be linked.
14.B.3 MRP logic

MRP is based on a very simple calculation. It looks at all the sources of demand
in each time period (‘time bucket’) and sums the total quantity required. It there-
fore gives no priority to any source of demand, since it expects to achieve the total
requirement. The basic calculation,
Initial Stock + Receipts – Demand = Closing Stock
is illustrated in Figure 14.6. If the inventory is sufficient to cover the demand in

that time bucket, then the same calculation is carried out for the next period. If the
inventory is insufficient, then delivery (order receipt) is required. It is then assumed
that this will arrive and this ‘projected available’ total is carried forward to find out
when the next delivery is required.
Starting with the current time period: if closing stock is above zero, the calcu-
lation is repeated for the next ‘time bucket’; if closing stock is below zero, then
create more receipts by ordering more. This creates extra supply for orders and
demand for components of that product. Including these planned supplies
enables the calculation to be repeated further into the future up to the ‘planning
horizon’.
MRP calculation for

each period
Existing opening
inventory on-hand 40
PLUS MINUS
Delivery receipts +10 Issues (demand) –30
EQUALS
Closing
inventory on-hand 20
Figure 14.6 Basic MRP calculation.
Example
A company made 30 chairs last week and had already five seats left
over from the week before. The demand was 20, so according to the
equation there were 15 left for this week. As they are making 30 again,
and customers want 37, there will be eight for next week, when the
supply of 30 will not meet the demand for 40, so extra components will
have to be ordered to manufacture the extra two.
This netting off process is carried out first for all periods for each customer
sales item. Where this is an assembly or manufactured product, the process is
repeated for the supplying level (components or assemblies). If this is also
composed of components, these are calculated at the next stage. The result is a file
(or a series of screens) showing:
• what has to be done by when;

• a supply schedule of when to confirm, place orders and receive deliveries;
• where appropriate, a manufacturing ‘work to’ programme for each production
area (called ‘production plans’ in Figure 14.5).
The MRP backward scheduling process starts by identifying the customer

requirement in the ‘master schedule’, which takes the customers’ demands
and creates a practical achievable plan for what will be done (not always to
the delight of every customer). How far ahead on the schedule will any
existing stock last (using Figure 14.6 successively)? When the inventory
runs out, deliveries have to be scheduled. If this is within the company, then
components will be needed. A bill of materials is required to cover this. At
the same time, information on how long the supply will take and where it
will be done – the lead times and routings – is necessary. All this informa-
tion is fed into the MRP calculation.
14.B.4 Bills of materials

Classic stock control deals with each item individually, independent of all others.
However, for most inventory, the usage of one item is linked to that of others (e.g.
printer cartridges and paper). Inventory controllers need a way to acknowledge
this link and to save the work of two forecasts where one will suffice (see
Figure 14.1). This can be done using an inventory bill of materials.
Traditionally, a bill of materials (BOM) is the list of parts, ingredients or
materials needed to produce or assemble the required end-product. The BOM for
a product is not simply a parts list; it contains more information. Most products
are packaged and so boxes, packaging materials, pallets and even documentation
is part of the full BOM. The BOM could also include consumables and tooling
replacements used in production; it is a useful way of planning for all the resources
which are necessary for the process.
In manufacturing, an assembly is made up of components. If they are all
bought-out and given to an operator to assemble and pack, then the BOM is just a
single level, as shown in Figure 14.7A. If there are several stages in processing
(e.g. subassembly, assembly separately), then the BOM is as shown in Figure 14.7B.
Inventory needs recording where there are delays between processes, and this
A. Flat BOM B. Structured BOM
Product Product
Assembly Assembly
Components
Subassembly
Components
Figure 14.7 Bills of materials.

requires structure in the BOM. In a supply chain, this corresponds to different
suppliers of items which are to be sold together in the end, whereas in manufacture
this is production through different processes.
14.B.5 Planning the structure of supply

Many activities result in the use of a variety of items, and if this usage is a fixed
ratio, or even if the ratio of usages is consistent on average, then the ratio can be
expressed in a loose BOM.
Suppliers providing a range of items to major customers can employ this
concept, especially where the causes of linked demand are understood. A plan-
ning (statistical) BOM can be created even where there is no manufacture
involved. Using these BOMs cuts down the amount of forecasting, since one
forecast will provide the information for all the items included in this bill of
materials.
Take advantage of links between requirements to co-ordinate supply.
The inventory control BOM usually consists of two levels, the same as a just-
in-time (JIT) manufacturing bill. The top level is the demand profile for the col-
lection of stock items for the customer. The level below is the dependent demand
for each of the items commonly making up this requirement. Since the relative
usage rates are not entirely fixed, the bill has to be monitored and corrected
according to the usage changes of the ratios. The supply chain situation is very
similar to a manufacturer where there is major variation in process yield or there
are quality rejects.
Example
A café providing hot drinks requires coffee, tea, milk and sugar, but
amounts required will vary according to the tastes of the clientele each
day. With experience, the ratios can be established and the supplies of
ingredients balanced so that there is always sufficient available, but not
too much of anything. In this case, the situation is trivial because of the
low costs and the short supply lead times, but the same principles apply
when an airliner is delayed for hours because one of the items required
to make a repair has to be flown in.
The demand for each item within the statistical BOM will vary for individual
customer orders. This means that actual safety stock of each item will consist of
two components:
1 Safety stock against the error in the overall forecast.

2 Individual line safety stock against the variances in the demand bill of
materials.
In practice, the statistical BOM is an excellent and under-utilised method for

distributors to give enhanced service where customers buy a range of items.
14.C Master planning

Within the ERP business planning system, the vital part for manufacturing and
supply chain is manufacturing resource planning (or MRPII). Failure to
understand this process has led to the demise of many companies. This was illus-
trated in Figure 14.5 and starts with the long-term plans, several years before the
inventory arrives. To be successful, the business has to decide on the market
focus and operations philosophy so that it can put in place the resources to
achieve it. These include people, skills profiles, plant, facilities, systems, partner-
ships, logistics. The outcome of this process is the ‘sales and operations plan’
(S&OP), which identifies demand ahead for at least six months and up to two or
three years, depending how fast-changing the market is. S&OP is usually based
on product groups rather than individual products and uses monthly or quarterly
time buckets.
14.D Master scheduling

Long-term planning sets the structure and constraints for the business, and at the
centre of the whole planning process is the master schedule. This was developed
to ensure the best availability for the customer within the constraints of the
operations and, as a result, improves the credibility of the company. A master
schedule is simply a list of what and when the business is going to send out to
customers (no excuses, no mitigating circumstances – those should have been
foreseen).
The master schedule decides the output for each product each week (or day in
fast-moving situations). It takes in the forecast demand from all sources
and matches it against the constraints such as material scarcity or processing
capability. Manufacturing capacity is always an issue, so the master schedule is
run though a ‘Rough Cut Capacity Plan’ to see if bottleneck processes can cope.
It is an improvement over classic stock control because it considers exactly when
demand will occur and the best way to fulfil it.
For example, if there is a known demand for 50 this week, 100 next week
and 60 the week after, a knowledgeable traditional controller could work
out the average demand as 70 per week with an MAD of 20 and then
decide on the appropriate safety stock. A different approach would be to
identify the demand in each week as 50, 100 and 60 and arrange supply
accordingly.
The master schedule is a forward-looking technique rather than historical.

The normal record of what has been issued each week is replaced by a plan of
what will be issued each week, showing individual weeks (or days) into the future.
This master schedule enables the planner to include weeks where the expected
demand for a product is very high, or the customer is on holiday. Figure 14.8 is a
simple layout for a master schedule. The quantities each week are the required
output after deducting the finished goods stock available for the customers.
The plan is arranged so that the total requirement does not exceed the supply
(or production) capacity.
The result of using a master schedule instead of a review level is that the
inventory is kept lower and large fluctuations in demand can be accommodated
within the process capability.
The master schedule is based on forecasts for stock items, so there is still a need
for some safety stock, calculated from the inaccuracy of forecast demand each
week (within the MRP, which results from the master schedule there should not be
safety stock). When customer orders are received, they convert the existing
forecast into firm demand. The master scheduler is responsible for arranging
customer demands into an achievable plan; this is management decision making,
not a clerical function.
Normally, particularly outside the demand lead times, the forecast quantity will
be greater than the total quantity demanded for each time period (a day or a week).
However, if orders are received which exceed the forecast, decisions are required
to deal with this.
Item Week 1 Week 2 Week 3 Week 4 Week 5 Week 6

Product 1 100 50 180 60 130 90
Product 2 30 45 45 30 35 60
Product 3 300 200 300 300 300 300
Product 4 25 25 25
Product 5 10 – – 10 – –
Product 6 – 100 – 50 – –
Figure 14.8 Simple master schedule layout.

Example
If the forecast is 100 per week and the first order is for 60 in the first
week, then it is no problem. If a second order is received for 30 in the
first week, it can also be supplied. However, if a third order is received
for 50, for that week, what are the options?
Working with the customer is essential.
The logic of allocation is, therefore, that the forecast is filled up by real customer
orders to the forecast level each week. However, if the customers require more,
then the master scheduler has to decide whether to:
• provide later delivery or part shipment to the customer (communicate this);

• reallocate goods already promised to another customer (risk bad reputation);
• overload the forecast and cause panic buying and inefficiency (high cost).
The last option is commonly used, but should be avoided since it makes supplier
and employee relations bad. Increasing capacity by overtime puts strain on the mate-
rial supply chain, which may not be able to react to short-term schedule changes.
Master planning takes the forecast and plans to supply parts to meet this demand
at the right time. In the more detailed master plan (Figure 14.9), the scheduled
Week Forecast Scheduled Projected Planned Projected Planned

requirement receipts (a) stock order available order
on-hand receipts release (b)
Current 50 50
1 300 500 250 250
2 400 500 350 350 500
3 200 150 150
4 200 500 450 450 500
5 400 50 50 500
6 300 –250 500 250
7 100 –350 150
8 300 –650 500 350
9 400 1,050 500 450
10 200 1,250 250
Current situation Planned situation
Figure 14.9 Master schedule supply planning.

Assuming: (a) a fixed batch size of 500; (b) supply lead time of 4 weeks; no safety stock.
receipts are the delivery date of current firm outstanding supply orders, and the
stock on-hand is the inventory which results if no action is taken. The first three
columns show the current situation. The remaining three show what has to be done
to meet the forecast demand (i.e. plan in new supply).
The planned order receipts stop the projected on-hand going negative, so mini-
mise stock and at the same time show how much and when to supply. The resultant
stock level is shown in the next column ‘Projected available’, i.e. the stock levels
resulting from these actions. With a four-week supply lead time, the order release
(last column) has to precede the order receipt by four weeks. If a planned order
release occurs in the current week, then an order needs placing now. If this is an
internal company work order, it will create demand at the next level down in the
BOM. Placing the order results in the quantity being deducted from the planned
order release and the ‘planned order receipt’ being transferred to a scheduled
receipt. If a safety stock is appropriate, instead of letting stock go down to zero,
then the ‘projected stock on-hand’ triggers a planned order receipt at the safety
stock level instead of zero. A safety stock of 100 would mean that stock would be
too low by the end of week 5 (Figure 14.9). The planned order receipt and release
would therefore be required a week earlier.
Set up supply using ‘projected available’ stock.
Allocate supply to customer orders using ‘available to promise’ stock.
14.D.1 Delivery promising

Once the supply side has been arranged and the ‘projected available’ planned, the
future inventory can be allocated to customers. The process of consuming the
forecast was discussed in Chapter 2 and it is shown in more detail in Figure 14.10.
The process is simply to estimate the demand and plan supply. The projected
available stock is then divided into the quantity pre-ordered by customer and that
which is unallocated.
This is ‘available to promise’, which should be an ideal tool for sales people to
give reliable delivery dates to customers. Companies that have poor adherence
to master schedules, and are continuously changing them, are not able to do
this. Better-managed businesses can use this, ideally as an app, during visits to
customers.
The calculations behind ‘available to promise’ are shown in Figure 14.11. These
are based on the projected available schedules from Figure 14.9. This is based on
the actual customers’ demand outstanding at the present moment, but orders are
constantly being received, so the situation may well change within a few minutes.
PLANNING Customer
Forecasts
SUPPLY schedules
Demand
Work already Avoid Creates new

started shortages planned supply
Projected available
Purpose: Sets up
manufacturing and
purchase orders
THEN
ALLOCATION TO
CUSTOMERS
Use
planned Sales order
Customer orders
supply promising
Unallocated supply
Purpose: Apply
customer orders
to the existing plan
Available to promise
Figure 14.10 Control of supply and demand.
This gives the ‘available’ quantity. Although demand totals 290 by the end of week 1,
the available to promise is only 10. This is because selling more than 10 before
week 3 would mean a shortage in that week and letting down customers, or chang-
ing the master plan. Disappointing customers loses business gradually; changing
the master plan, even if it is possible, leads to reactive management, high costs and
poor performance. Therefore, keeping within ‘available to promise’ is a key busi-
ness constraint.
Delivery Customer Scheduled Stock Planned Available Available

promising orders receipts on-hand order to promise
receipts
Current 50 50 10
1 260 500 290 290 10
2 660 500 130 130 10
3 120 10 10 10
4 320 500 190 190 70
5 120 70 70 70
6 60 10 500 510 110
7 400 –390 110 110
8 10 –400 500 600 600
9 –400 500 1,100 1,100
10 –400 1,100 1,100
Figure 14.11 Available to promise.

Scheduled and planned order receipts from Figure 14.9.
The master schedule shows what is available for customers and when.
14.D.2 Batch sizes

The master schedule is an excellent tool for managing and monitoring stocks, espe-
cially where there is variable demand which can be estimated. It enables lower stock
to be held than can be achieved by statistical stock-centred methods. It also links in
very well with supplier scheduling. Getting the best out of master planning does
require more effort, because of the regular reviews and re-forecasting. However, it
can be of good benefit, especially for A class items and those where the demand is
understandable and variable (e.g. customers taking large batch quantities at infre-
quent but regular intervals). The triggering of MRP by master scheduling enables
companies, particularly manufacturers, to provision for whole ranges of stock with
a single evaluation and give a balance of stock for customers who require co-ordi-
nated supply (in classic MRP, a number of batch rules have been developed as many
of these are based on ‘economic batch sizes’, which produce batch quantities that
are unbalanced and too large for customer order requirements).
The same logic can be used for dependent demand order quantities as for nor-
mal stock control. The three main alternatives are:
1 lot-for-lot;
2 fixed batch size;
3 period cover.
1 Lot-for-lot is the first choice because it leads to a minimum of inventory. Get
what is required during each time bucket.
The rule is to bring in the balance required in each time period. In Figure 14.9
the scheduled receipts for weeks 1, 2, 3 and 4 would have been 250 (300
forecast demand less the 50 stock), 400, 200 and 200. The projected on-hand
would always be zero unless there is a policy to hold a finished goods safety
stock which then becomes the minimum instead of zero.
This gives a variable supply quantity, which is inconvenient in some
circumstances.
2 Fixed batch sizes are used widely to illustrate MRP (see Figure 14.9). They
are overused in practice since changes in demand level should be reflected in
changes in order quantities. Fixed batch sizes are convenient to use, but are
only essential where there is a restriction in vessel size (distribution or
manufacturing). Batch sizes based on price breaks are the result of too
simplistic an approach to costing.
Choose batch sizes to minimise work and inventory.
3 Period cover is a way of taking the demand for the next few periods and
batching it up. It is the same as lot-for-lot, but with several time periods added
together. Again, the order quantity is variable, and the stock level is signifi-
cantly higher. This technique reduces the number of orders and increases their
size at the expense of higher inventory. The application is therefore to C class
items where this is the most beneficial option.
If five weeks’ worth were supplied in weeks 1 and 6 (Figure 12.9), then the
period cover would be 1,500 – 50 = 1,450 and 1,300.
Key points
• Look for structure in demand patterns.
• Working with dependent demand keeps inventory low.
• MRP is a simple way to trigger future supply orders – especially
where order levels vary.
• Using ‘available to promise’ gives more accurate delivery dates.
• Batch sizes have to be defined using rules which minimise total
costs and workload.
15 Supply chain inventory
management
• Effective supply chains.

• Basic requirements.
• Avoiding the Forrester effect.
• Where to hold inventory – the 9-box model.
• Distribution chain management.
• VMI benefits and consignment risks.
• Replenishment systems – fair shares and DRP.
15.A The basis of the lean supply chain
15.A.1 Overall inventory management

The science of inventory management has developed from the statistical analysis
determining stores inventory levels to supply chain optimisation. Inventory man-
agement is no longer a single problem, but it involves organising for maximum
efficiency with suppliers and customers. The manager is now co-ordinating:
• warehouse operations;
• inventory control;
• distribution and logistics;
• suppliers;
• customers;
and attempting to optimise the profit for the supply chain rather than just improve
their delivery-on-time while reducing inventory. However, the targets are still the
same, although in a wider context:
• better customer service;

• lower inventory;
• decreased operating costs;
218 Supply chain inventory management
but now for a lean supply chain, not on the individual echelon. In theory, all infor-
mation can be available immediately (‘big data’), although some companies are
reticent to share. Good inventory management can give a critical competitive
advantage. The challenge is now:
• collecting data accurately;

• being willing to share it along the supply chain;
• selecting the appropriate data;
• understanding how to use it most beneficially;
• working for the good of the supply chain and not for political gain.
The supply chain needs to co-ordinate flow, batch sizes and information using the
techniques discussed in previous chapters.
15.A.2 Lean supply

The principles discussed in Chapter 6, lean supply, are applied throughout the
supply chain. In brief, these were:
• customer focus – exceeding the customers’ expectations;

• aim for perfection – not just better than the competition;
• added value – only carry out activities which directly benefit the customer;
• flow – so that inventory does not stop, it continues from one stage to the next;
• unit movement – small batches (e.g. kanbans) moved frequently;
• simplicity – in operation so that things are obvious and do not go wrong;
• individual responsibility – at operator level for developing more effective
processes;
• integrated design approach – with additional features for specific customers
and ease of manufacture at low cost;
• quality – always create the correct quality first time and do not pass on defects.
Use lean supply ideas to see how to develop inventory control and logistics.
Improvement is a circular process of tightening up one area then moving onto

the next and eventually back again to the original as it has become the weakest
area again (the same as tightening on a cylinder head in a car engine). Efficient
supply chains require focus through all the businesses on:
• goods delivered on time consistently, fit for purpose, labelled and packaged

correctly;
• design, so that a small number of products satisfies a wide range of customers
and are composed of a small number of components;
• continuous reassessment of processes, designs and operating structures.
Supply chain inventory management 219
In the supply chain, the end user demand is small, often sporadic, but the aggre-
gation of demand back along the supply chain and the use of generic supplies
smooths the demand if managed correctly. Forecasting is therefore more accurate
further back in the supply chain (unless the Forrester effect takes over).
15.A.3 Information accuracy

All the work on inventory management will come to nothing if the base
information is wrong. There are many companies where the quality of informa-
tion on the system is very poor. If the same quality standards were applied to
product quality, then they would quickly be out of business. Part of the develop-
ment of better inventory control is therefore to ensure that the information on the
formal system is:
• real-time;
• accurate.
More accurate records reduce inventory holding.
Systems are accurate, so it is a matter of failsafing the input and making

people believe that it should be correct. Inaccurate records have a high cost,
arising from continuously checking to see if items exist and holding extra inven-
tory to avoid stock-outs, not to mention the obvious costs when customers can-
not have the items promised to them. It is the responsibility of warehouse
management to provide accurate information. Stores have two types of custom-
ers, those requiring physical stock and those requiring accurate information, and
they are equally important.
15.B Co-ordination
15.B.1 Global supply

There will always be somewhere in the world where manufacture is cheap, and
people need supporting. International trade has always provided items which were
only available from certain countries because of climate or minerology. Mostly, it
is now caused by low labour costs. International bulk supply causes long lead
times for many products. Many companies have mistakenly simply considered
unit cost as the criterion for outsourcing. They should have considered cash flow,
since unit cost includes a significant proportion of fixed costs (management, mar-
keting, operations and accounting) which remain and other costs (procurement,
warehousing, financing inventory, freight forwarding) are increased significantly.
However, global supply enables a consistent product to be offered throughout the
world, giving efficiency in design, operations, marketing and volume. Extended
lead times require better forecasting, especially where the product lifetime is short
(e.g. mobile phones and electronics). It is important, therefore, to co-ordinate the
stockholding throughout the supply pipeline. The option of high-cost air transport
instead of slow sea containers reduces the risks from unforeseen demand to be
covered.
As low-cost manufacturing companies embrace more lean practices, supply
may reduce batch sizes and provide variety flow.
15.B.2 Forrester effect

Soon after the development of computers, Jay Forrester was modelling a dynamic
supply chain and identified major problems with its operation due to batch-
ing and lead times (it is also known as the ‘bullwhip effect’). An illustration will
help.
A fairy story
A customer buys one item from a retailer, who then needs to re-order and so
places an order on the wholesaler for a pack of 12. This takes the wholesaler’s
inventory below the review level, so they order replenishment of 250 from
the central warehouse. This depletes the warehouse stock, so they order
another batch of 3,000 from the manufacturer. The manufacturer, using
traditional thinking, makes 10,000 and orders the materials for a further
10,000 from their suppliers. The actual demand was one. How much inven-
tory is remaining in the supply system and how long will it take to use it up
if the demand is typically one per week?
This illustrates part of the problem; the other part results from the supply
arriving too late, as follows.
A second tale
A company sells items at 20 per week and gets them on a six-week lead time
from their supplier. However, the demand increases to 35 and 40, so the
company has to order extra because the variance and the forecast have now
increased. However, the demand in the following four weeks is back to
about 20. After the original six weeks, the extra inventory arrives, but it is
not needed, so the supply orders are delayed until the stock decreases to the
normal level, when new supplies are again ordered. The supplier therefore
sees high demand followed by no demand followed by medium demand:
the smoothly varying demand gets broken up and becomes more extreme
further back in the supply chain. The only compensating factor is that demand
becomes more generic further back in the supply chain, so the fluctuations
from each supply channel can even each other out.
Back along the supply chain, fluctuations in the demand amplify and, although the
decisions at each point appear sensible, the overall situation is stupid. Unfortunately,
most supply chains suffer from the Forrester effect. The resultant problems have
caused closure of efficient manufacturing companies in the author’s experience.
The problem is avoidable. Ideally, the answer is kanban. Unfortunately, this
requires immediate supply, which does not happen all along the supply chain but
should be employed wherever possible to link processes (especially in manufactur-
ing). Where companies are dispersed and selling to each other on a lead time, then
an integrated approach is necessary. Would the manufacturers have made 10,000 if
they knew the demand was only one per week? The answer depends on whether
they are looking for short-term profit (ignoring the 7,000 they made which they
probably won’t sell). They would definitely not have ordered any more materials.
The answer, therefore, is to understand the demand and eliminate uncertainty.
In practice, this means:
• integrating the information for the supply chain. Enable all companies to see
the inventory and usage data.
• measuring end customer demand. Collect and share sales data leaving the
pipeline.
• reducing the number of stages. Integrate stores data systems, change
warehouses to cross-docking locations.
• agreeing management of the inventory. Appoint a controller for all the
inventories in the supply chain.
• reducing the lead time. Take out inventory duplication and improve flexibility.
• structuring the inventory. Work together to agree best inventory strategy.
The solutions are easy, but getting businesses to co-operate in their implementa-
tion is more difficult.
15.B.3 Demand driven MRP (DDMRP)

MRP systems were always over-optimistic, expecting the master schedule to
remain correct over extended lead times. The materials planned did not fulfil the
actual demand because the master schedule had been updated and practitioners
had to make ad hoc arrangements. Acknowledging this, DDMRP (demand driven
MRP) was developed to enable the risk to be reduced. By incorporating nodes of
safety stock at strategic points into the MRP structure, the integrity of the
MPR calculation is restored. This enables a controlled level of inventory to be
introduced where necessary to cover the fluctuations in demand.
15.C Supply chain operations
15.C.1 Objectives
Supply chains are a relatively recent concept because they go against our natural
instinct to ‘go for a bargain’. Gradually, businesses have learned that price is not
as important as quality and reliability, and for repetitive requirements a longer-
term arrangement reduces overall costs. Nowadays, inventory managers have to
maintain and balance supply partnerships, distributed stocks, additional inven-
tories, distribution costs and manufacturers’ work-in-progress with selected
customers and suppliers. They have to apply their skills to ensure that the supply
chain is:
• customer service oriented throughout;

• integrated (in operation and information);
• consistent;
• flowing (minimum warehouse time);
• reliable (quality delivery and information).
Optimise the complete inventory structure, not each stage.
15.C.2 Supply chain structure

Inventory needs to be located strategically in the supply chain so that it meets the
expectations of the end customer and at the same time maintains economic oper-
ations in terms of operating costs and capital investment. Ideally, the end users
would like to have inventory locally so that there is no risk and immediate avail-
ability, but, as this is often costly, suppliers prefer more central locations. The
competing attributes are shown in Figure 15.1.
Some general principles should be considered when deciding how to structure
the supply chain. Normally, the challenges are in divergent supply chains where
the advice is to ‘put inventory as far back in the supply chain as possible while
maintaining the delivery and integrity expectations of the end customer’, while
also considering that:
• fast delivery costs more;

• the greater the distance, the greater the cost;
• distribution costs are generally higher for smaller loads; and
• regular shipments cost less than one-offs;
• distribution costs are largely loading and unloading, mainly independent of
distance;
• consolidation saves cost and is easier for small, regular deliveries;
• small loads need smaller transport and facilities, which are cheaper.
Stock in a single central warehouse Stock held in many local stores

Minimum investment in stock Items available rapidly for customers
Accurate forecasting Sensitive to local conditions
Higher demand rate Duplication of stockholding
Reduced safety stocks Poorer control
Low warehousing costs Higher inventory investment
Better formal physical control Greater risk of obsolescence
Accurate integrated system required Higher operating costs
High distribution cost to end customer Low distribution cost to end customer
Low transport cost to central warehouse High total transfer costs to local stores
Figure 15.1 Inventory location options.
The complexity arising from the latter four considerations makes transport
costing difficult. Just-in-time supply (Chapter 6) appears to increase distribution
cost, but through continuous improvement (kaizen) this may be ameliorated.
To have a significant effect on the cost balance, bold leaps have to be made in
distribution structure and warehouse location. The size and frequency of shipment
has to be altered dramatically. This usually means a change in means of transport
to provide smaller units faster. The savings in inventory and warehousing costs
should be greater than any marginal increase in distribution costs.
The whole inventory system can be considered as a pipeline for supplying the fluid
‘stock’ to the point of sale (or use). The input to the pipeline is a continuous (sched-
uled) flow. For very-high-usage items, the pipeline can continue to the point of use.
For other fast-moving goods, a central warehouse can act as a junction diverting
replenishment to local storage tanks directly without any delay or storage. The cen-
tral store is used purely as a transhipment point for these items. The pipeline is then
connected to the distributed stocks. The central inventory management function
simply co-ordinates supply scheduling and ensures consistent delivery performance.
For low-usage items, the movement includes a reservoir, central stores, from
which the items are dispensed as required.
15.C.3 The 9-box model

The distribution of stock along a supply chain providing products to many users
depends upon the:
• urgency of supply required by the end user;

• geographic dispersion of warehouses;
• inventory value;
• costs of distribution;
• warehousing costs;
• degree of co-operation along the supply chain.
Low valve Medium High valve

fast valve fast fast
movers movers movers
Annual demand
αc αb αa

medium valve medium
movers medium movers
βc movers βb βa

slow valve slow slow
movers movers movers
χc χb χa
Unit cost
Figure 15.2 The 9-box model.
They are specific to a particular supply chain and business. Shaping the inven-
tory enables these complex situations to be managed more easily. ABC analysis
(see Chapter 3) is the first step. However, in ABC classification, C class includes
both high-volume low-value items and slow-moving high-value items. For a single
warehouse, the C class supply policy is the same: buying ten weeks of demand
(maybe costing £100) for a slow mover may be just one unit, whereas for a fast
mover worth £0.10 it could be 1,000 items. In a distribution supply chain, the pol-
icy would be different. The fast-moving items can flow along the supply chain to
near the end users, whereas to minimise the total inventory of high value C items
it is better to hold it centrally and incur the higher cost of fast distribution. In the
9-box model (Figure 15.2), the unit cost (abc) and demand rate (αβχ) are consid-
ered separately, and have separate strategies.
Pareto Analysis determines how to control items at one stage of the supply
chain. The 9-box model uses the same principle to develop this approach or to
decide where to keep items in the supply chain. Pareto Analysis is focused on
achieving an availability by structuring supply, whereas 9-box analysis also sculp-
tures the availability. As a general concept:
• αa products generate most of the revenue;

• χa products sell at a premium, but providers need to be well organised;
• αc products are often low margin commodities where efficiency is key;
• supermarkets and cost cutters will avoid providing χc goods.
Companies can decide on the focus of their business to suit their target custom-
ers and therefore structure their offering. This is primarily organised to provide
items to customers within an acceptable timescale from a designated stage in the
supply chain. The business can then produce balanced plans like the one illustrated
in Figure 15.3, which is the result of taking the likely contribution to customer
satisfaction, profit created and operational capabilities.
Hold stock Maintain Daily

at end user small POS replenishment
interface stock to POS
Annual demand
Central or Stock at
Offer on
distributed distribution
lead time
stockholding centre
All time
Small stock
overbuy Make-to-
held
distribute as order
centrally
required
Unit cost
POS = point of sale to end user
Figure 15.3 9-box structure for supply.
991/2% at 95% at 99% stock

user user & fast
interface interface supply
Annual demand
αc αb αa
99% from 97% from 95% from

supply supply supply
chain chain chain
βc βb βa
Patch 100% 85% from Wait lead

or make-to- central time, make-
order stock to-order
χc χb χa
Unit cost
Figure 15.4 Availability structure.
There are then two aspects to optimise:
1 At what stage in the supply chain should the inventory be held (assuming that
duplication has been eliminated)?
2 How much inventory is required at this stage?
The first aspect is purely a cost balance taking account of the relative ware-
housing and logistic costs between sources and end destinations. Companies that
own sites can use these economically, but renting premises enables locations to
gradually be moved as the customer base changes. The second aspect can be con-
verted to a matrix of availability KPIs which define the safety stock for each box.
An example is shown in Figure 15.4. The relative availability values will be
selected to meet the needs of the particular market being served.
Local management Central management

Advantages Advantages
Sensitivity to local conditions Good usage statistics
Customer knowledge High level of control
Local commitment Centre of excellence
Responsible for success Allocate stock to greatest need
Pragmatic commercial arrangements Professional inventory expertise
Disadvantages Disadvantages
No overall view of inventory Lack of local knowledge
Expertise required at each location Remote from customer problems
Potential for hoarding Highly effective systems required
Different local policy implementation Risk of incorrect local information
Requires inventory control specialists
Figure 15.5 Central and local management of inventory.
15.C.4 Management of inventory

Assessment of demand is more accurate for the total demand than for individual
warehouses, since the larger volume has a relatively smaller MAD (the MAD only
increases as the square root of the volume). This means that historical forecasting
is most accurate when carried out by a central planner. However, if demand
patterns change (new large customer, loss of contract, etc.), then the local know-
ledge of these events needs to be included in the forecast (see Chapter 11).
Local managers of inventory usually have other expertise and a variety of
responsibilities, of which inventory level is a part. For inventory planners
controlling stock in several regions, their only function is to maintain supply and
minimise the total operating costs and inventory investment by holding the right
amount of stock in the right locations. They lose out on the local knowledge and
interaction with customers, but are able to use specialist expertise and better
statistical information to make decisions. The balance is made by considering
the topics in Figure 15.5. The better option often depends on the quality of the
communications and the ability of the planners.
15.D Replenishment techniques
15.D.1 The stock-keeping unit (SKU)

The term SKU is used widely and often incorrectly. It is not the same as ‘item
identity or product code’ because it not only identifies a unique item but also its
value in a specific location. The value is a combination of product cost and logistic
cost. The SKU has a unique cost for the item at each site, and provides the infor-
mation required to optimise distribution.
Use the power of SKU costing to make informed decisions.
Example
A chair is made in China and bought by a European business for £10. It
is shipped to Europe. An individual in India sees it advertised and buys
it for £25 (including delivery). Did the European business make a profit
and, if so, how much? It is not possible to tell because the logistic cost
is not known. The cost at each stage should include the transfer cost,
so that informed decisions can be made. This is included in the defin-
ition of an SKU.
If the European business had known that the shipping cost was £8 in
both cases, then it is unlikely that they would have carried out that sale.
However, if the shipping cost from China to India were also £8, then the
business could have offered the chain via this route and made a profit
(although it may not have been ex-stock).
The proper definition of SKU is therefore ‘the identity code of a unique type of
item at a specified location with an attributable total cost’. Obviously, the logistic
cost should include the normal distribution cost and, where companies cannot iden-
tify this properly, the same SKU code should be used across several locations.
Another example
A customer in Newcastle, UK, orders a whole lorry load of beams. This
is a major order for the Sunderland branch, which is 14 miles away.
It uses almost all their stock, so they get some more from the manu-
facturing plant in Newport, South Wales (296 miles away). As a result,
lorries are loaded and unloaded twice and extra return mileage was
done (28 miles). As long as Sunderland get the credit, direct shipment
would have been cheaper. If a planner was trying to minimise costs or
maximise profit, and knew that the SKU value at Sunderland was higher,
then the right decision would have been more likely.
15.D.2 Vendor managed inventory and consignment stock

Sharing responsibility for stock along the supply chain can reduce the overall
inventory, reduce total lead time and improve flexibility. Where a company does
Vendor Customer
Who owns the Customer managed raw material
inventory inventory stock
Supplier on- Consignment

Supplier
site stock stock
Supplier Customer
Who manages the inventory
Figure 15.6 Supply inventory options.
need to store stock from a supplier, there may be four alternatives, depending
whether they control the stock or own it (by paying for it). These options are
shown in Figure 15.6, the ‘customer’ being the business that has the stores and
needs the stock on-site.
Option 1: vendor managed inventory (VMI). An appropriate level of inventory
is agreed with the supplier, who then provides a top-up to this level on a regular
basis. The customer pays for the goods in the normal fashion. This is an excellent
technique because it frees the customer from having to monitor and order, and it
provides the supplier with ongoing sales. It enables the supplier to plan replenish-
ment logistics and frequencies for maximum efficiency.
Examples
The manager became overloaded with purchase requests when the
health and safety regulations required a large variety of extra handling
consumables to be used. Although these items are C class and bulk
purchase would alleviate the problem, there still remained the need to
order and avoid stock-outs. By negotiating VMI for this wide range of
items with a single vendor, the manager’s work was eliminated, the
stocks were always available and the value of inventory was kept low
(and in this case the contracts reduced the total purchase price).
Another (manufacturing) company uses plastic granules for all its
products. These are delivered into large silos linked to the process
machines. These silos have load cells which measure the weight of
material. These are linked to the system at the supplier who can then
monitor usage and send deliveries either when the stock gets low, or
when it is most economic to deliver.
VMI uses the expertise of the supplier and there is no need for daily monitoring
and management by the customer. In many instances, the lower inventory enables
delivery to the point of use instead of stores. For the supplier, there will be
contracted regular consistent demand. The risks for the customer are that the
supplier may over-deliver to improve sales and thereby inflate stockholding. This
can be avoided quite simply by looking at the invoices and ensuring that there is
not gradual unjustified inflation. Suppliers also have to ensure that the contract is
viable: they need to be aware of demand changes and advise the customer, or else
they will be subject to sporadic demand and returns.
Option 2: customer raw material stock. This is the normal traditional arrangement
for suppliers and customers as discussed in the previous chapters. If the demand is
‘dependent’, ideally the supplier should deliver just-in-time for the usage; however,
because of lead times and uncertainty, this is a stockholding node in DDMRP.
Option 3: supplier on-site stock. The customer does not have the cost of the
inventory if the supplier is on-site or next door (a good situation for kanbans).
Service vending machines are an example of this and, where the business is high
volume, the suppliers can set up adjacent to the customers or vice versa. Again,
this is normal commercial arrangement, just with short lead times.
• Ensure supply, avoiding transport or production delays.

• Avoid capital cost.
• Transfer the supply responsibility to the supplier.
• Enable plans to be changed without panic buying.
• Avoid worry by the customer.
Option 4: consignment stock. This is inventory owned by a supplier at the cus-

tomer’s site and has, mistakenly, been popular. Customers like consignment stocks
because they:
• avoid capital cost of inventory;

• ensure supply, avoiding transport or production delays;
• transfer the supply responsibility to the supplier;
• enable plans to be changed without panic buying;
• avoid worry by the customer.
Suppliers agreed because:
• on-site inventory effectively captured the market;

• the customer should not have shortages;
• delivery was arranged when convenient, not on short lead times;
• it ensures continuing business.
However, consignment stock increases supply chain costs and reduce competi-
tiveness. In practice, consignment stock will:
• not reduce the cost, but just transfer it and the customer pays eventually;
• usually lead to high stock levels in the consignment because both customer
and supplier sales like it;
• require improved recording accuracy and stores discipline;
• lead to disputes over unrecorded usages;
• make it difficult to operate the phasing in and out of products;
• cause duplicate back-up stock in suppliers’ stores;
• take the pressure off efficient supply.
As suppliers’ consignment is stock in a remote location it is generally poorly man-

aged and anyway is not available to service other customers. It adds to the cost of
running the supply chain.
Consignment inventory is only a temporary fix.
The consignment stock should not be palliative to poor planning. Where there
is an essential but unreliable supplier, there is a reason for holding inventory, and,
since the cause is the supplier, it would be sensible for them to bear the cost until
they can resolve the problem. The buffering of supply in the short term (few months)
can be a solution to cover a specific problem.
Simple rules to use when arranging consignments are:
• Have a project plan for how the problem will be resolved with a completion
date for removing the consignment.
• Do not set stock levels; set availability targets instead.
• Review variety of inventory and levels regularly (every few weeks).
• Use consignments only for a few fast-moving item types.
• Arrange who manages excess inventory caused by design changes, or user
demand, etc.
• Allow consignments to be available for other users.
• Have a project agreement with financial sanctions.
Of the four options, consignments are the worst option, and minimising
stockholding for one of the other three assists the smooth running of the supply
chain.
15.E Managing distribution

The techniques commonly used for managing the distribution channel are:
1 Distribution requirements planning (DRP; for planning supply).

2 Fair shares (for allocating available stock).
Both techniques are concerned with the quantity of stock to be shipped through
the distribution network.
15.E.1 Distribution requirements planning (DRP)
DRP collects demands from distributed stores and consolidates them into a
time-phased plan for despatches from a central warehouse.
Each store at the user end of the supply chain generates a forecast for demand
which is used to identify the orders they are likely to place on the central ware-
house. The forecast from each store covers much more than the supply lead time,
so that a longer-term supply schedule can be created. DRP is an extension of the
MRP process for distributed warehouses. The simple calculation and logic used
for MRP is used in DRP (see section 14.B.3). The process is:
• Future customer requirements are identified in each local store by week

or day.
• Starting with the current period, the demands are netted off against the
available stock.
• If the remaining stock is above the safety stock, then this quantity is available
for the next time period and the calculation is repeated to the end of the
forecast.
• If the available stock is less than the safety stock, then a delivery is required.
The quantity requested depends on the batch rules being used.
• The delivery generates a supply schedule or order a lead time earlier. The lead
time corresponds to the picking, packing, transfer and receiving time.
• Assuming that this will be delivered, the calculation is repeated to give the
full delivery schedule into each local store.
As these local stores are fed from a central warehouse, the demand pattern on
this warehouse has now been forecast. This is illustrated in Figure 15.7. Here,
three stores receive goods from a central warehouse. The delivery lead time for
North and West is one week, whereas it takes two weeks for South stores (the lead
time includes the demand processing time and the weekly delivery). The current
stock, the expected sales demand and the result of the safety stock calculations are
shown. From these, the projected total demand by the stores on the central supply
warehouse are calculated each week (note that the requirement of 17 for the South
stores should have been sent in the current week, so is likely to be late unless
expedited). The current 190 stock in the central warehouse (not including the 17)
is then used for the stores, from which is calculated the timing of the external
supply (with fixed batch size of 380).
DRP relies upon the stock controllers in each store to give a forecast of demand
which they will be supplied with. This does not allow for alterations in their
forecasts outside their safety stocks. This arrangement can work well where
the end customer has an accurate agreed schedule (which some automotive
manufacturers can do). DRP feeds seamlessly into manufacturers’ master
schedules. It looks at demand far ahead of the normal lead times and produces a
time-phased shipping requirement for the suppliers which is also efficient for
transport arrangements.
Location Activity Safety stock Current Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7
stock
North stores Demand 75 50 45 60 20 50 60 45

20
Requirement 0 40 60 20 50 60 45
West stores Demand 15 20 30 0 50 20 30 15

12
Requirement 17 30 0 50 20 30 15
Supply 1 week lead time North stores 17 70 60 70 70 90 60
South stores Demand 300 100 140 60 110 150 140 135
46
Requirement 0 0 46 110 150 140 135
Supply 2 weeks lead time South stores 0 46 110 150 140 135
Central supply Total demand from 17 116 170 220 210 225
warehouse stores
Warehouse stock 37 190 74 284 64 234 389
Deliveries planned from supplier 380 380 380
Figure 15.7 DRP stock and forecast demand.

Note: Safety stock is a calculation used to trigger replenishment not physical stock.
Like MRP, the supply chain is planned by backward scheduling from the
customer. This requires the establishment of fixed supply lead times and batch
quantities. Manufacturers traditionally like large batch sizes, which means that
intervals between deliveries are correspondingly long. This causes high stock in
the supply chain and so a continuous replenishment approach with small batches
is a great advantage.
Some pitfalls which should be avoided, because DRP is less flexible, are through:
• focus on individual stores rather than the total system;

• concern with achieving stores available stock not overall customer service;
• fixing and depending upon the supply performance.
Wherever possible, the manufacturing and distribution systems should be

synchronised.
DRP relies on the store inventory planning at the customer interface to be
accurate. This is difficult for them to do because the demand is relatively small and
there may be business reasons for over or under estimating the usage. The tech-
nique is relatively rigid, since the requested quantities must be sent on the days
scheduled to the destination originally planned.
The DRP stores replenishment quantities are planned well ahead, so changes to
shipment schedules have to be long term rather than immediate. There is little
possibility of responding to major shortages since the quantities shown are
planned to be available, not currently available. DRP can result in considerable
unevenness of the gross requirements, caused by aggregating the independent
forecasts from the stores; it is best suited to situations of relatively steady demand.
It works well where fixed delivery quantities are physically necessary (such as
container loads).
DRP is a useful way to plan products into the distribution chain.
Production master schedules’ quantities and time phasing can be created dir-
ectly from DRP, since they supply the process. The batches of 380 in Figure 15.7
are the container quantity coming straight from the manufacturing site. This elim-
inates the need for finished goods stock. Since the supply is planned to meet the
requirement well ahead, the demand rate may have changed, so the resultant
replenishment delivery may be too early or late by the time demand is met.
15.E.2 Fair shares distribution

Although DRP is a useful way of identifying requirements into the distribution
chain, when the products arrive, the situation could have altered and the anticipated
demands changed. The balance of stock among the stores is likely to be different.
How the central supply warehouse reacts will depend on who has the responsibility
for the inventory and their requirements. There are three basic situations:
1 The local stores manager has forecast and ordered the items, wants them and
will pay; in which case, the goods are sent out per the DRP plan.
2 The local stores manager has over-forecast and so does not really want extra.
3 The central supply planners control the stockholding.
In the third case, and sometimes in the second, stock arriving at the central
distribution warehouse can be sent to the region of greatest need – some will get
more than they asked for and others less. The strategy is to maximise the stock
availability across all the stores, and maybe upset some stores’ stock controllers at
times. With fair shares analysis, there is no need to hold inventory in a central
warehouse since distributing it to the local stores makes it available for the cus-
tomers. An implicit assumption is that, within the supply network, lead times are
relatively short, otherwise the planning aspects of DRP have to be used.
When stock is delivered into a distributed stores network, it is reallocated to the
various stores in the most equitable manner. The sensible approach is to cover
priorities first and then share out the rest to meet future demand. Fair shares analy-
sis is a method for achieving this using a standard procedure. The logical process
is to allocate supply to shortages and safety stocks in the local stores and, where
necessary, in the central supplying warehouse and then to allocate the remaining
available stock to meet the requirements in proportion to the demand rate.
The fair shares allocation of incoming stock normally distributes stock in the
following sequence:
1 customer back orders unsatisfied due to shortages;

2 expected demand within supply lead time (anticipated shortages);
3 safety stock at distributed stores;
4 any safety stock at the central supplying warehouse (none if organised well);
5 allocation of remaining stock in proportion to demands at each local store;
6 for high-value and excess stockholding, redistribution where a stores stock
exceeds its share (if the logistic cost justifies it).
If stock is insufficient to meet the full requirement at any stage, it is allocated

according to this logic. The actual stock transfers used to achieve this distribution
can be optimised through the transport algorithm. A simple example of the use of
the fair shares logic is shown in Figures 15.8 and 15.9 where the 380 delivery in
week 2 of Figure 15.7 is allocated by fair shares analysis instead of DRP. The total
stock quantity at the start of week 2 is recorded in Figure 15.8 and then the pro-
cesses 1 to 5 are followed in Figure 15.9 (process 6 was unnecessary in this case).
The results are not rounded quantities, instead they reflect the new demand
situation at the time of supply into the central warehouse. This is particularly use-
ful when there is significant lead time between ordering and delivery (or inflexible
schedules) and when the forecasts do not come to fruition.
Inventory start of week 2
Location Quantity
Warehouse stock 74
Deliveries from supplier 380
North stores (Current stock - week 1 demand) 25
West stores Shortage of 5 0
South stores (Current stock - week 1 demand) 200
Total inventory available 679
Figure 15.8 Inventory available for fair shares.
Location Arrears Safety LT stock 3Wk3 Allocation Total Current Transfer

stock requirement stock
North stores 20 60 20 40 140 25 115
West stores 5 12 0 50 16 83 0 83
South stores 46 110 150 113 419 200 219
Total 5 78 170 220 169 642 225 417
Central 37
Warehouse
Figure 15.9 Fair shares allocation.
It is assumed for simplicity in these figures that the demand and safety stock
quantities have remained the same during week 1, which is normally not the case.
Also, the calculation has left safety stock in the central supply warehouse, which
could be allocated if there is confidence that there will be no major forecasting
problems at the stores.
Fair shares works when the central planners have the power to direct the inven-
tory to the point of most need. As the forecast statistics at the centre are more
accurate than at the local stores, it is generally better to use these forecasts and
then fair shares when the items arrive (with the potential to use the centre purely
for cross-docking and no stockholding). The local knowledge with the DRP
approach adds value, so this could work well where is significant regional variation
and close communication with end users.
Fair shares analysis approaches the regional stock issue as an inventory control
problem, assuming that distribution costs are not the prime concern. The distribu-
tion quantities calculated by the forecasting techniques and apportioned by fair
shares often have to be adjusted because of other influences. The total shipment
may have to be phased because of space limitation in particular stores. The
distribution quantities could well be rounded to the nearest pack size, or altered to
take advantage of a distributor’s cost structure.
15.F Recycling and reverse logistics
As people look for total efficiency improvements, two concepts can enhance each
other:
1 re-use;
2 waste transport capacity.
Over the long term, cost can be minimised by returning items to suppliers and
making them available for use again. This concept ranges from the recycling of
materials to sales on internet sites (such as eBay), where customers buy a variety
of alternative items and return those not required. Recycling is supported by legal
requirements (e.g. WEEE) and international waste agreements. In practice,
recycling includes:
• re-usable, repairable modules;

• packaging materials, pallets and stillages;
• end of life equipment;
• basic materials (e.g. paper, metals, plastics, glass).
The basic premise is that the cost of recycling should be less than the cost of
disposal and buying new. The ‘cost’ can be considered in several ways, ranging
from the unit cost to the user, to the environmental cost including waste manage-
ment, social impact and long-term economics.
A key element in the operation and costing of recycling is the design of the item
initially. Has the design taken this into account? New product designers need to
consider the selection of materials, the creation of modular and products that can
be dis-assembled, and the costs over the lifetime of the item. Purchasing decisions
for many types of equipment are now based on ‘lifetime costing’ – consisting of
the purchase and installation costs, the operating maintenance costs incurred
while in use and, significantly, the decommissioning and sale cost. Waste disposal
is becoming increasingly expensive, so design for re-use or recycling is important.
15.F.1 Reverse logistics

Reverse logistics is considered to cover the transfer of goods back down the supply
chain towards the source. This operation is often a major headache of inventory
management, since the customer considers the return of these goods as a low
priority because they are defective or old and so have low value. However, these
items are important supplies for the vendor, especially in the case of:
• pallets, stillages and packing. Some customers are dilatory in returning

specialist, often expensive, items which are essential for the transfer of prod-
ucts. Sometimes, the same customer will not accept goods without these and
deliveries can be delayed.
• repairable modules, which will be needed to fix equipment failures in the future.
In these cases, extra inventory has to be purchased if the reverse flow is slow,
delayed or accumulated into batches before returning.
If ‘T’ is the average time for repairable modules or pallets to progress around
the cycle or ‘repair loop’ (i.e. the average time taken getting round the loop to the
same place in the same state) and the usage rate of the item is R, then the minimum
number of units required is R × T. As the supply (especially the reverse flow)
fluctuates, a safety stock has to be added to this. The safety stock is calculated in
the normal manner. It should be used for the whole loop and held strategically at
the place of most impact; if safety stocks were held at other places in the loop,
then this would be unnecessary excess stocks. A full view of inventory at all stages
of the repair loop enables inventory to be minimised.
Key points
• When the inventory in a warehouse has been improved, consider
the supply chain as a whole.
• Minimise the logistic costs while exceeding customer expectations.
• Decide upon the best places to situate the stock and the amount of
stock to hold in each, using inventory management principles.
• Operate the supply using planning tools like DRP and, particularly,
fair shares.
• Control of the whole repair loop has a major effect on inventory
investment in recycling items.
16 Meeting the challenges
• The process for improvement.

• Communication.
• Information sharing.
• Technical advancement.
• Making the process work.
16.A Review
The development of inventory management is a dynamic activity, with new
approaches being made, techniques being refined and new challenges being met.
Successful inventory management is a balance between the use of basic
techniques, which are easy to use and give reasonable results, and sophisti-
cated techniques, which can give better control if they are used properly, and
co-ordination of supply and demand. Current management is not limited by
technology, information storage or processing power, it is limited by the will-
ingness to provide information, to ensure that it is correct and to avoid instanta-
neous over-reaction. Often, businesses suffer because they do not realise the
central role of inventory management and therefore do not foster the required
expertise. Continuing competitiveness is based on ever improving customer ser-
vice and ever reducing inventory costs, requiring greater expertise in inventory
management.
The science of inventory management as developed in this book has to be
carried out by an organised, aware and dedicated artist. The manager has to focus
on the simple basic requirements of creating the most profit for the organisation
through the best customer service. This means:
• excellent forecasting;
• low inventory value;
• small operating costs;
• co-ordination with suppliers and customers.
Meeting the challenges 239
The direction is toward the impossible dream – perfect service, no costs and no
inventory. To manage this, it is essential to have operating controls and targets –
monitoring of current performance and tighter targets (which are achievable and
agreed with those required to reach them).
There is also the balance between long-term and short-term requirements.
Paying extra delivery costs to help the customer may be good for relations. As a
one-off it is fine, but as a frequent activity it can damage the viability of the
business. The inventory controller has always to look simultaneously at the long
term and the short term, detailed picture and overall plan. By steering a consistent
course, based on a set of operating principles, work is reduced and effectiveness
increased. As better control is introduced and the performance improves in one
area, so it can be integrated with other activities and new challenges adopted.
Take a simple example of a manufacturer that used to hold raw materials inven-
tory and then supply the finished items into a warehouse from where it was distrib-
uted to regional stores. Each stage of the process was struggling to gain good
control in its own separate way.
Some people are still at base 1 and wrestling with these challenges. There is a
natural process of evolution from this setup:
1 Gain control over individual stores and manufacturers (through master

scheduling and MRP, for example).
2 Include finished warehouse stock to the manufacturing control process (from
materials management activity).
3 Add distribution costs into the cost balance (identify SKUs).
4 Form partnerships with suppliers to avoid duplicating their stocks with raw
material stockholding.
5 Work on flow through the supply pipeline to eliminate finished warehouse or
distributor stocks.
6 Integrate the distributed regional stores into the inventory management
system (DDMRP approach).
7 Integrate SKU records along the supply chain.
8 Co-ordinate supply chain management to minimise supply chain costs
(‘integrated business planning’).
9 Continuous development of product flow.
Companies should identify which stage they have currently reached. In some
cases, participants cannot see that they benefit, so the process is difficult. However, it
is necessary if businesses are to remain competitive, because other companies are
developing and improving, so maintaining the present situation is not a viable option.
A company that is operating in the same way it was five years ago
has a problem.
From a base 150 years ago, when the storeman was the inventory controller,
science has developed through quantitative stock level assessment to linked inven-
tory structures. The development of inventory management is now proceeding in
diverse directions, which can be considered.
16.B Collaboration
There has been a culture change in inventory management. Instead of developing
better quantitative tools, the focus of inventory management has changed from the
technical aspects to interaction, flow and co-ordination: it is the aspects of working
together, communication and data availability where improvements have been
significant. It is a long time since purchasing was viewed as a competitive struggle
between supplier and customer. With repetitive supply, everyone realises that all
the companies need to make life easier for the others, so that the end customer
receives the variety they want, quickly and at low cost.
Good relationships enable this to happen, so the development of a good match
at company and interpersonal level is required, with the development of improved
approaches or templates based on CRM for effective supply chain interaction. The
challenge is how to provide good, accurate, timely information. The more demand
can be measured and understood, the lower the inventory required to fulfil it.
Therefore, there has been much development of techniques for collaboration.
Operation of the supply chain is by people, so working together is key.
Strategists are attempting to pull these together under umbrellas such as

‘collaborative planning, forecasting and replenishment’ (CPFR), which gives
structure to the co-ordination of pipeline activities to avoid detailed sensible
decisions becoming stupid overall planning (as seen through the Forrester effect).
It uses the techniques in earlier chapters in a supply chain context, and provides a
framework in which to motivate people to improve operations. CPFR is about
sharing information in a structured manner to minimise the overall operating costs
and enables vertical integration for independent businesses along the supply
chain.
16.C Information structures
16.C.1 Enterprise resource planning (ERP)

The advent of interactive databases enabled the interaction of information from all
areas of business and groups of companies. ERP systems aimed to provide the
complete information base for business, and is successful, especially, in integrat-
ing the accounting processes. The potential in the areas of planning inventory and
production still has to be realised. The original dedicated systems dealt with the
detail, but this functionality is often lacking in ERP systems and some planners
still resort to spreadsheets.
However, important features are now routine, especially sales and operations
planning (S&OP), which makes top management consider providing the resources
(equipment, systems, infrastructure, suppliers, personnel and skills) sufficiently
far ahead to make it available when needed to meet the master schedule. Also, the
use of optimisers, such as advanced planning and scheduling (APS, or sometimes
APO), assists decision making.
16.C.2 Big data

Where operational data is made available from companies throughout the supply
chain, computer analysis can provide insight and automate decision making.
The inventory manager’s capability for optimising and co-ordinating inventory is
much enhanced by big data analysis techniques and devising suitable algorithms.
To achieve this in real time requires feedback. This has led to development of
internet-based monitoring of many activities (‘Internet of Things’). The use of this
voluminous data collected from diverse sources, in real time, is a major break-
through. Not only do managers have warnings and solutions, but they can then
control processes remotely (the same as an app controlling central heating or air
conditioning from the mobile phone). The application of this to the whole supply
chain has obvious advantages for response to customer requirements and flexibility
and puts pressure on the manufacturers for agility, providing small quantities at low
cost. To make these basic concepts work successfully, there has to be a great deal
of trust, co-operation and unified goals for the businesses within the supply chain.
16.C.3 Supply chain activities

Inventory is kept to near zero by JIT kanban suppliers, but many businesses obtain
items from far away on long lead time, often in big batches, and therefore have
warehouse inventory. The decision is primarily a financial one and, on that basis,
supply sources will gradually alter as countries prosper and others become less
expensive. The economics of local, global or in-house should be compared through
the potential cash flow, not through item ‘profit’.
Where there is lead time there is risk, so the development of DDMRP enables
the whole supply chain to be structured so that the ‘nodes’ of stock at strategic
stages absorb the changes in demand and MRP can work without continuous
revisions and panics. Many businesses can trim their inventory and solidify their
planning by organising with their suppliers and customers. These are the first steps
in embracing the move toward big data.
An aspect that has to be addressed is supply chain resilience. This can be
considered as reliability and risk mitigation; in other words, making sure that
things can’t go wrong and, if they do, how supply performance can be protected
should it happen. Risk analysis for personnel protection is highly developed and
the same diligent strategies can be applied to maintaining the demand chain.
16.C.4 Single warehouse inventory levels
The optimisation of warehouse inventory was key to much mathematical modelling
last century, partly based on ‘orders’ and the discredited EOQ model. The focus of
development has not been on improved mathematical models; instead, work is
based on lean supply and supply chain co-ordination. There is opportunity to
improve the normal, simple calculations and provide better availability per unit
inventory investment.
The use of simulation for solving individual situations provides valid answers,
but it is difficult to create general principles from this approach. Developments
integrating review level, supplier delivery frequencies and 9-box logic could
improve KPIs, if the mathematics prove easy to apply.
16.C.5 Forecasting
Time series historical forecasting can be very complex. Improved models give
benefit, but understanding of demand, collection of end-user data and modelling
of causes of requirements are proving more useful.
There is potential for better forecasting and inventory management using
fuzzy logic, but currently this requires mathematical modelling expertise, and
those who have used it so far have not attacked the basic inventory management
problem.
16.C.6 Customer focus

As customers want solutions to their situation, suppliers that take responsibility
for doing this are the successful ones. Providing items from inventory is not as
lucrative as providing a service, so good companies are giving individual offerings
to suit each customer. The benefits from customer satisfaction and retention
outweigh the added costs. The effect on inventory is an increase in variety and the
provision of unique items to customers.
Servitisation includes these concepts as well as the care for their offering
throughout its lifetime. This includes maximising the lifetime benefits and
minimising the total lifetime costs, comprising purchase, installation, running
costs, disposal and all associated costs (such as infrastructure). Further
developments of concepts will assist customers in their decision making and
control.
16.C.7 Additive manufacturing

Will many supply chains be redundant? If items can be reproduced at point-of-
use, then storage and distribution will no longer be required. There is potential for
these production processes to be employed by the end user or at a local facility
from a downloaded design. This will depend on the benefits (speed, cost, design
advantages) compared to conventional supply. The specific advantages of additive
manufacture are therefore:
• single item production;

• potential for product designs not makeable conventionally;
• no lead time (assuming materials are available);
• zero stock;
• no supply chain.
The first established additive technique was 3D printing, making physical

products by building up layers of material. There are two stages to the process:
transferring the design to a printer readable format, and then using the printing
machine to create the item. As techniques develop for depositing more diverse
materials, wider uses will be possible.
Other additive techniques are now becoming available and are developing for
situations where mechanical strength is required. Production by ‘fused deposition’
requires more advanced equipment, but could solve problems of providing items
in remote locations where availability is paramount.
16.C.8 Understanding finance

One of the issues facing inventory management is economics. The conventional
accounting processes do not provide a good basis for decision making. Internal
business costs are a mixture of fixed cost and variable cost and it is often difficult
to unravel the two. The cost of holding stock is mainly a mixture of the warehouse
costs (usually fixed), the cost of shifting goods (people, often fixed) and inventory
finance. To make a decision on expanding the warehouse, outsourcing to a third
party or taking on extra staff, is not well served by the accounts. What should be
considered is the actual cash implications of the development: how much extra
money is spent and how much is saved. Often companies forget that most of
the overheads, particularly management costs, are unaffected or increased by
outsourcing and operating costs are fixed unless extra personnel are required.
Conventional accounting is fine for informing stakeholders of the long-term
health of the business, but careful consideration of the real costs and benefits, or
the use of techniques like ‘throughput accounting’ and ‘discounted cash flow’ are
necessary for decision making. Future developments of inventory strategy should
be based on sound economic facts.
16.D Recipe for success

The key to success is simplicity of operation. The concepts, systems and structure
can be complicated, just as reality is, and the inventory management system has to
reflect this. It does not mean that the actual day-to-day operation should be
difficult. In fact, the more sophisticated the system, the more reliable it should
be and the less effort to operate. However, a good system requires setting up to
operate properly and this requires good inventory management expertise, the
professional techniques described in this book. Using the forecasting described
here, along with ABC analysis and stock level calculations, the author has saved
millions of pounds’ worth of inventory in many companies. Most businesses want
to generate more profits; many have not recognised the key role of inventory
management in increasing turnover and reducing costs.
Inventory management is a collection of techniques developed by practical people
and proven in the real world. It is no use using general good management practice in
this area where there are specific inventory control methods required. These have
taken many years to develop and refine. Most practitioners are lagging behind best
current practice and work with techniques that give reasonable outcomes.
It has been known for decades that Newton’s law of gravity is not quite right
and Einstein’s theory of relativity is nearer the truth, though not perfect, but
Newton’s law is good enough for those of us who are not space travellers or
who move below the speed of light. Similarly, in inventory control, the
review level calculations are not exactly correct and delivery frequency
does affect the safety stock.
Aim to get it nearly correct now rather than perfect sometime in the future.
Better formulae are readily available, but, from observation, stocks are so far
out from the ideal that a small amendment to the safety stock level is not going
to alter the real inventory. Similarly, Gaussian statistics, simple availability
calculations and independent demands are all concepts which are questionable
theoretically, but which work in practice.
More advanced forecasting methods provide better answers when the future can
be relied upon to mirror history (see Chapter 13). Crude tools like ABC analysis
could be replaced by tools which take into account the potential different stock
policies for high-value slow-movers and low-value fast-movers. Or the 9-box
model could be developed into a differential value analysis, which gives an indi-
vidual class for each item code and optimises every inventory line. This evaluation
will take place when companies have exhausted the current methods and realise
that they can benefit.
The concepts discussed in this book have been tried and tested to give compan-
ies major improvements in their efficiency across a wide variety and different sizes
of business and industries. Understanding the full implications and applications is
a gradual process, but worth the effort because of the tremendous benefits.
This book only takes a snapshot of the situation. The reader can access all these
topics on the internet. On the basis of the understanding here, hopefully the body
of knowledge being created will improve the best practice in our profession.
Key points
• Embrace new ideas.
• Use the techniques within the supply chain.
• Communication is key.
• Develop new technology.
• Focus on the basic objectives.
Questions and answers
Questions
Q1 What are the three key objectives of inventory control?
Q2 Which departments put pressure on inventory managers, and what are their
objectives?
Q3 ‘More stock gives better availability.’ Challenge this conventional view.
Q4 Two companies distribute fast-moving consumer goods. Their results for
last year were:
Table Q4
Annual accounts data Ace Distributors, £million Deuce Associates, £million
Sales turnover 8 26
Fixed assets 2 5
Cost of sales 6.2 21
Stock value 4 8
You have been offered the job of inventory manager by both companies.
Which one would you take? Based on the data provided, explain why.
Q5 The aim for Ace Distributors (in question 4) is to reduce the stock to
£1.5 million. What will be the return on assets then?
Q6 What change in the return on capital (as a percentage) will there be when a
company has reduced its stock value by £500,000?
Sales turnover = £3 million

Profit = £400,000
Capital plant = £1.1 million
Stock before reduction = £1.5 million
(Capital employed = stock + capital plant)
Q7 Discuss the ways in which the JIT approach differs from conventional
inventory management.
Q8 How does a company give good customer service?
A company is distributing a range of fast-moving items to stockists. The
following is a list of despatches for one day. What is their customer service
level, assuming this to be a normal day?
Table Q8
Order No 1A Quantity requested by customer Quantity despatched from stores
Item No 1 20 15
Item No 2 100 100
Item No 3 50 20
Order No 2B Quantity requested by customer Quantity despatched from stores
Item No 1 60 40
Item No 2 150 100
Item No 3 20 20
Order No 3C Quantity requested by customer Quantity despatched from stores
Item No 1 10 10
Item No 2 70 70
Item No 3 50 20
Item No 4 25 25
Q9 Which items in the following list are in the A, B and C classes?
Table Q9
Item code Weekly demand Unit price
A901 2 £3.00
B662 2 £8.00
C355 20 £4.00
D523 40 £0.50
E191 10 £65.00
F807 1 £9.00
G010 3 £1.00
H244 1 £170.00
J459 10 £3.50
K488 22 £0.50
Q10 If the current stock of the items in question 9 were as shown below,
Table Q10
Item Stock quantity
A901 6
B662 10
C355 50
248 Questions and answers
Item Stock quantity
D523 60
E191 50
F807 50
G010 12
H244 5
J459 20
K488 5
a) what is the stock cover for each of the items?

b) what would be the first task resulting from this analysis?
c) what is the stock cover value for the total stock?
Q11 A stores is offering ex-stock delivery to a wide variety of customers. Item
35721 is a non-seasonal item which has the following usage characteristics:
Weekly usage 200

MAD 50
Supply lead time 4 weeks
a) If deliveries are brought in every week, what is the target stock level if
85% availability ex-stock is required?
b) If the supplier is persuaded to deliver daily, what would be the new target
stock level?
c) How would the change affect the value of stockholding if the items cost
£10 each?
Q12 The statistics on the stock of a range of items are as follows:

Table Q12
Item no Unit cost Annual usage
2A32 £25.00 12
2B44 £0.50 360
3D10 £0.10 120
3E82 £40.00 1
3F66 £2.00 40
4G19 £5.00 520
4H95 £4.00 7
5J53 £20.00 5
7N78 £0.30 200
9P21 £12.00 50
The stores controller is under pressure from the management to reduce the
stock level to below half a month’s usage. The stores personnel are adamant
that they are not going to do any extra work. Show what the stock policy
should be in these conditions and how the target is achieved (assume that
safety stock is not necessary in this case).
Q13 Stock levels are as follows:
A items £28,000 130 items
B items £6,500 350 items
C items £3,500 600 items
Obsolete items £2,000 200 items
Within the budget for next year, the estimated usage rate for each is:
A items £133,000 per month

B items £16,000 per month
C items £2,000 per month
Obsolete items Nil
What is the best way to reduce the stock by £2,000?

Q14 Do you believe that lean supply simply passes the burden of holding stock
on to the supplier?
Q15 Your managing director has asked for a short report on ways of reducing
work-in-progress and throughput time. At present, batch sizes are about 15,
set-ups average three hours, processing time per unit is two minutes, and
WIP stands at 13 weeks. Create a short report outlining in simple language
the measures you would take.
Q16 Kanbans
a) What are the financial benefits to a company of setting up a linear flow
kanban system?
b) A company is intending to set up a kanban system for production.
Comment on the changes which will be required in:
i) Responsibility;
ii) Layout;
iii) Operator training;
iv) Quality management.
Q17 What is the role of safety stock, and why is it often not required in a JIT
environment?
Q18 ‘Lean supply is a good theoretical idea, but cannot work in our company’,
said the production director. ‘We have customers who change their minds,
a variety of processes, many products, and real people who cannot always
be relied upon.’
Set out the arguments which you would make to persuade this director that
JIT is a good idea for the company.
Q19 The data below is the stock record for item 345.
Table Q19
Date Receipts Unit cost Issues Balance
14/3 10 11 10
12/4 3 7
22/5 2 5
10/6 5 15 10
30/6 3 7
17/7 2 5
a) What is the value of the last issue (17/7) of item 345, based on:
i) FIFO costing;
ii) Average costing;
iii) LIFO costing.
b) If you needed to reduce the stock value in your stores by £500,000,
what items would you look at first and what techniques would you use
to reduce the stock value?
Q20 Why is safety stock held?
Q21 What are the three factors which determine safety stock?
Q22 The usage for two items over the past five weeks has been:
Table Q22
Item Week 1 Week 2 Week 3 Week 4 Week 5
D523 60 30 20 40 50
K488 30 15 0 65 0
a) What are the values of average demand and variability for each?
b) The items cost 50p each and the lead time from the supplier is two
weeks. The inventory controller is deciding whether to try and provide
a service level of 90% or 95%. What is the additional cost of the better
service?
Q23 ABC Distributors have the UK dealership for a well-known manufacturer
of pre-packaged confectionery. Their main activity is the supply of goods
to the local retailers ex-stock.
At the recent stocktake, ABC Distributors had 1,280 different stock lines
with a total value of £40,000.
They order all their items during the week they fall below the re-order
level.
The system works well except for the fact that the stores manager and his
assistant are overloaded with work – in fact they are currently three weeks
behind with the ordering. As they are ordering one month’s requirement at
a time, there are often times when the stock runs out. The suppliers take
two months to deliver, and the shortfalls cannot therefore be rectified
immediately.
What are the changes in operating practice that you would suggest to assist
them? What controls should they use which will bring the changes into
effect?
Q24 ‘Re-ordering systems are not designed to stop you running out of stock.’
Give two reasons why this statement can be justified.
Q25 What factors do you consider when setting stock levels for an independent
demand item?
Q26 The demand history of item XX1 is as follows:
Table Q26
Week Demand
1 60
2 50
3 70
4 60
5 80
6 70
The items are supplied weekly in boxes of 100. Work out:
a) Review level;
b) Average stockholding.
(You can assume a safety factor of 2 for 94% customer service.)
Q27 Explain why, for some A items, we can expect the review level to always be
higher than the physical stock level?
Q28 A company is putting expensive fountain pens into boxes at a rate of about
250 per week, together with ink cartridges. The stock controller has decided
that the pens they buy are A class, while the cartridges and boxes are
C class.
Suggest order and delivery patterns and order quantities for the three types
of item, and discuss why you have chosen them.
Q29
a) Calculate the review level for the supply of umbrellas, given the
following data:
Lead time = nine weeks
Customer service level expected is 90%
Demand over the last six weeks has been 11, 23, 37, 24, 10 and 15.
b) Discuss the assumptions which you have made in using this calculation.
Q30 A company is proposing to schedule in supplies of 20 key items each week
instead of ordering them monthly.
a) What preparatory work do they need to do with the suppliers?
b) If the purchase value of these items is £1 million per year, by how
much will the average inventory be reduced?
c) What are the benefits and extra jobs caused by scheduling?
Q31 What factors determine the delivery quantity?
Q32 Discuss ways in which stock can be reduced by suppliers and customers
managing a supply chain together. What are the pre-requisites for success-
ful supply partnerships?
Q33 What is the difference between a target stock level system and a review
level system?
Q34 Forecasting
Weekly sales of a product are:
Table Q34
Week Demand
1 12
2 14
3 8
4 16
5 13
6 18
a) Discuss what forecasting techniques can be used to determine the

demand for the next week. Indicate which you would prefer to use and
the reasons for your choice.
b) If the initial forecast for week 1 was 10, work out the forecast demand
for each week using exponential smoothing, and give your prediction for
week 7. You can use an α factor of 0.2 for this calculation.
Q35 What items of data (inputs) are required for the MRP calculation? For each,
state the level and type of accuracy necessary to ensure reasonable results.
Q36 A watch manufacturer has a schedule for assembly covering ten weeks. The
schedule is:
Table Q36
Watch X Manufacturing orders for 1,000 watches due in weeks 2, 4, 6, 8 and 10.
Assembly lead time per batch is one week.
Watch Y Manufacturing orders for 500 watches due in weeks 3, 6 and 9.
Assembly lead time per batch is two weeks due to a special processing
on the watch cover.
Watch Z Manufacturing orders for 100 watches due every week.
Assembly lead time is half a week. The parts for the batch of week 1
have already been issued.
Watches X and Y use the same module, A, and batteries (two per watch).
Watch Z uses a different module, B, and the same batteries, but only one per
watch. Batteries and modules come in boxes of 500. There are seven boxes
of batteries and three boxes of modules in stock. The buyer has outstanding
purchasing orders for a total of 20,000 batteries and 6,000 of each module,
half quantities to arrive in weeks 4 and 10.
a) What would you expect your MRP package to recommend for

purchasing? (Assume that the package is not capable of recommending
re-scheduling.)
b) What would you recommend should be done?
Q37 A product is manufactured in two stages. The material for the process consists
mainly of one basic, bought-out ingredient. Expected sales of the product are
as follows:
Table Q37
Week 1 2 3 4 5 6 7 8 9 10
Sales 210 300 140 65 200 260 150 100 190 220
The stock situation is:
Raw material 500

Semi-finished 220
Finished goods 650
The manufacturing and supply situation is:

Operation Lead time
Purchasing three weeks
Bulk production one week
Finishing two weeks
a) What should the weekly purchase plan be on a lot-for-lot system?

b) What should it be using with a 500 batch size?
Q38 Distribution planning
a) What different underlying assumptions are there between DRP and fair
shares allocation of goods?
b) What is the fair shares priority logic for distributing stocks to regional
warehouses?
Q39 Prepare a schedule for a supplier, using the DRP technique, for the follow-
ing forecast demands and lot size replenishment quantities.
Table Q39
Lot size Period demand Current

stock
1 2 3 4 5 6 7
London 600 150 320 370 190 150 320 370 600
Manchester 250 70 70 70 70 70 70 70 230
Newcastle 400 90 150 150 90 90 150 150 310
Internal sales Monthly 30 30 30 30 30 30 30
call-off
Lot size for central stores is 1,000. Current stock for central stores is 1,650.
Replenished lead time for all local warehouses is two periods; internal sales
are replenished immediately; supplier lead time is three weeks.
Q40 What are the advantages and disadvantages of keeping stock in retail stores
rather than in supply depots?
Q41 The following ten items are the range of stock held in a small stores.
Table Q41
Item no Unit cost (£) Annual usage
1 150 2
2 125 100
3 85 10
4 50 70
5 40 1
6 26 20
7 20 3
8 10 1
9 7 300
10 6 20
a) Identify which are the A, B and C class items.

b) Discuss how ABC is used in purchase ordering.
c) Review another use of this Pareto classification.
Q42 An item is delivered weekly to a business on a one-week lead time. The
company has a firm demand, shown in the table.
Table Q42
Week 1 2 3 4 5
Demand 10 20 5 10 15
What is the average quantity held in stock to meet the demand for this item,
treating it as:
a) Independent demand (assuming that a 99.2% customer service level is
required)?
b) Dependent MRP demand (level 2)?
c) If the company can arrange a daily kanban supply for this firm demand
and the items are despatched at the end of each day, what would be the
average inventory? Specify any assumptions you make.
Q43 It has been remarked that ‘stock level for an item does not depend on
its lead time’. Defend this remark using examples, and discuss why the
statement is not precisely true.
Q44 The central warehouse has to supply three depots with stock, but there is a
shortage. The stock in each of the depots and the calculated safety stocks
and demand levels are given in Table Q44:
Table Q44
Depot Current Safety Overdue Demand Demand

stock stock depot sales this week next week
Dangley 7 10 0 10 12
Brassby 0 5 2 8 10
Carbock 20 15 0 12 18
There is none in the central warehouse, which supplies these three depots
for a whole week at a time. A consignment of 55 arrives at the central ware-
house. If all is to be sent out to the depots:
a) Explain the fair shares priority logic for distributing stocks to regional
warehouses.
b) Calculate the number of items to be sent to each depot.
Q45 On taking up your new job in an established business supplying a number
of products ex-stock, you discover that the company does not have any
proper demand forecasting except a guess by the purchasing manager, who
buys from experience. You have been asked to outline what the company
should do to create a formal forecasting process. Briefly describe your
recommendations.
Q46 Midas Manufacturing Ltd is an independent company with 200 employees
and is well-established in its marketplace with its specialised range of prod-
ucts. Two different customers have each individually approached Midas
about going into a partnership with them. Discuss the risks and the merits
of Midas choosing each of the following as partners:
i) A small, fast-expanding family company in the UK, currently having
20 employees.
ii) A multinational group selling a wide range of products with a centralised
sales office in South Africa.
Q47 The actual sales data for a textile is shown in Table Q47. Starting with a
forecast of 54 for week 1, calculate the forecast for each week, using α = 0.3,
and predict the demand for week 8.
Table Q47
Week (t) Actual demand (metres) (A)
1 55.3
2 54.5
3 57.8
4 55.0
5 57.2
6 61.2
7 55.6
Answers
As inventory management is not an exact science, some latitude is possible in the
answers. The important aspect is whether you have understood the concepts.
1 – an answer
Availability (delivery on time); minimum inventory cost; minimum operating
cost.
See Chapter 1.
2 – an answer
Purchasing, if they try to buy large batches to reduce unit cost.
Sales, if they want large stocks to service possible demand.
Accounts, who require lowest investment in inventory to improve cash flow.
In manufacturing, Production traditionally want to make large batches and to
have buffer stock in case of problems. Also pressure comes from elsewhere –
designers who specify items too late and management who keep changing
the targets.
See Chapter 1.
3 – an answer
High stocks result from poor control, and poor service results from poor man-
agement of supply and forecasts. The better the inventory management, the
better the availability and the less stock cover is required.
4 – an answer
Table R4
Annual accounts data Ace Distributors £million Deuce Associates £million
Annual profit* 1.8 5

Return on sales 1.8 5
8 26
= 22.5% 19.2%
Assets are 2+4 2+4
Return on assets 1.8 5
6 13
= 30.0% 38.5%
*
Sales turnover: cost of sales.
Ace has better margin and more potential for improvement in stock control.
5 – an answer
51.4%
Table R5
Annual accounts data Ace Distributors £million
Annual profit 1.8

Return on sales 1.8/8 = 22.5%
New value of assets 2 + 1.5 = 3.5
Return on assets 1.8/3.5 = 51.4%
6 – an answer
Capital employed before reduction = stock + capital plant
= £1.5 million + £1.1 million
= £ 2.6 million
Capital employed after reduction = £ 2.6 million – £500,000
= £ 2.1 million
Return on capital before reduction = £400,000/£2,600,000
= 15%
Return on capital after reduction = £400,000/£2,100,000
= 19%
7 – an answer
There are two key ways in which lean is a different approach:
1 The level of inventory is not a simple calculation; it is a negotiated
quantity.
2 Lean does not set a permanent inventory situation; it empowers the people
to improve and reduce the levels required.
This means that those involved with managing the supply chain always
have to see ways of improving reliability of performance, reducing times-
cales and costs and therefore removing the need for inventory.
The timescales of kanban operations are measured in hours, under the
umbrella of a long-term relationship and smooth supply. Conventional
inventory management works with days or weeks of lead time and inven-
tory levels and any demand pattern.
8 – an answer
Solving customer requirements by providing quality goods, service, informa-
tion and ongoing support, and by delivery on time.
Table R8
Demand satisfied on time: Despatched Requested
By order 0 3 0%
By order line 5 10 50%
By item code:
Item 1 65 90 72%
Item 2 270 320 84%
Item 3 60 120 50%
Item 4 25 25 100%
Average % fill per line 77%
Totals
Average item availability 420 555 76%
Different answers are used in different situations. As the customers hold inven-
tory, the demand for each item is separate, so measuring the availability by
order (at 0%) is not appropriate. Also, as the supply will top up customers’
existing stock levels, full delivery of each line is desirable but not essential,
as long as the balance is supplied before they run out (so measurement by
order line (50%) is a better measure, but not the best.
Average item availability (measured against demand at 76%) could be bet-
ter, but could bias the result toward the high volume (and often low value)
items. Therefore, for this situation, average % fill per line (at 77%) is favoured.
This enables a business to provide more inventory to the customers with most
need (fair shares).
9 – an answer
Table R9
Item code Weekly demand Unit price Turnover Rank Class
A901 2 £3.00 £6 9 C
B662 2 £8.00 £16 6 C
C355 20 £4.00 £80 3 B
D523 40 £0.50 £20 5 C
E191 10 £65.00 £650 1 A
F807 1 £9.00 £9 8 C
G010 3 £1.00 £3 10 C
H244 1 £170.00 £170 2 B
J459 10 £3.50 £35 4 C
K488 22 £0.50 £11 7 C
See also Chapter 3.

10 – an answer
a) The stock cover would be:
Table R10A
Item Stock Usage Cover
A901 6 2 3.0
B662 10 2 5.0
C355 50 20 2.5
D523 60 40 1.5
E191 50 10 5.0
F807 50 1 50.0
G010 12 3 4.0
H244 5 1 5.0
J459 20 10 2.0
K488 5 22 0.23
b) Ensure that supplies of item K488 are arriving within one day.
c) This is calculated on a financial basis.
Table R10B
Stock Usage Unit value Stock value Turnover
A901 6 2 £3.00 £18 £6

B662 10 2 £8.00 £80 £16
C355 50 20 £4.00 £200 £80
Stock Usage Unit value Stock value Turnover
D523 60 40 £0.50 £30 £20

E191 50 10 £65.00 £3,250 £650
F807 50 1 £9.00 £450 £9
G010 12 3 £1.00 £12 £3
H244 5 1 £170.00 £850 £170
J459 20 10 £3.50 £70 £35
K488 5 22 £0.50 £3 £11
£4,963 £1,000
Stock cover for all stock:
= £4,963/£1,000
= 4.96 weeks.
11 – an answer
a) TSL = demand in lead time and review period + safety stock in LT and RP
= 200 × (4 + 1) + 1.3 × 50 × (Square root (4 + 1))
= 1,000 + 145.3
= 1145.3.
b) Assuming that the lead time is still four weeks and orders are placed daily:
TSL = 200 × (4 + 0.2) + 1.3 × 50 × (Square root (4 + 0.2))
= 840 + 133.2
= 973.2.
c) This would be a reduction of 172.1 in TSL
The average stock is TSL – half usage in (LT + RP)
So, reduction would be from 645.3 to 553.2
= reduction 92.1 units at £10 each = £921.
12 – an answer
There are several ways of getting to this answer. The easiest is to perform
Pareto Analysis and then apply month, and C items every four months (except
those with too low usage rates; as shown in the table).
Table R12
Item no Unit Annual Usage Class No of Average Percentage

cost (£) usage value per orders stock of value
year (£) per year value (£)
2A32 25.00 12 300 B 12 12.50 7.5%

2B44 0.50 360 180 C 3 30.00 4.5%
3D10 0.10 120 12 C 3 2.00 0.3%
3E82 40.00 1 40 C 1 20.00 1.0%
3F66 2.00 40 80 C 3 13.33 2.0%
4G19 5.00 520 2,600 A 52 25.00 65.0%
4H95 4.00 7 28 C 3 4.67 0.7%
5J53 20.00 5 100 C 3 16.67 2.5%
7N78 0.30 200 60 C 3 10.00 1.5%
9P21 12.00 50 600 B 12 25.00 15.0%
Total 4,000 95 159.17
This gives about 95 deliveries instead of the 120 if every item is ordered each
month. The total cost of sales is £4,000 per year (£333 per month) and the
average stock will be about £159, which is 47% of the monthly demand as
required. See Chapter 3.
13 – an answer
Table R13
Class Stock (£) Budget (£) Stock cover % reduction required

(weeks) for £2,000
A 28,000 133,000 0.88 7%

B 6,500 6,500 1.71 or 31%
C 3,500 2,000 7.35 or 57%
Obsolete 2,000 0 Infinite or 100%
Although the stock cover for A class is already under a week, it should be
easier to gain £2,000 by reducing this because it represents the majority
of stock value. This can be achieved by scheduling or kanban. In practice,
some inventory reduction may also come from the other classes. See
Chapter 4.
14 – an answer
It does not pass on inventory; it eliminates it. Suppliers have to be able to
respond rapidly and deliver the smooth demand pattern required (see Chapter 6).
Kanban supply is consistent and therefore suppliers require little safety stock
even if they provision in batches themselves.
15 – an answer
From the figures, the average batch only takes 30 minutes to produce, but
takes three hours to change over. This is inefficient. There needs to be a
concerted effort on reducing the set-up time. At the same time, product
design could be reviewed to see whether common parts can be created so that
set-ups can be avoided. There is also a problem with high WIP. Looking at
the wasted time in between operations will show where this wasted time can
be eliminated. The use of a kanban pull system would cause the WIP to decrease
dramatically. See Chapter 6.
16 – an answer
a) The costs of manufacture and stockholding are reduced. Kanban creates
an environment where the WIP of the A class items is less than a day’s
worth between each process. The processes are linked together so that
there is a flow of work. This enables efficient manufacture and a reduc-
tion in waste, which contribute to lower operating costs.
The improvement in speed through the process means that any reject
work is caught earlier and therefore can be rectified before too much
material and capacity has been used up. There is also the direct reduction
in inventory, which removes working capital costs.
b) Responsibility: Shop floor responsible for triggering supply of next item/
batch without fail. Supply operation responsible for replenishing in time.
Layout: Processes arranged in sequence with minimum distance between
the processes.
c) Operator training: Operators responsible for their area and processes
within it. Teamworking skills required as well as multiskilling. Operators
need to be able to transfer from one process to the next in case there is a
bottleneck.
Quality management: Excellent quality is a precursor to introducing
kanban. Quality should be built into the process so that it will not go
wrong. Quality is part of the manufacturing process, not a separate
activity.
17 – an answer
Safety stock decouples one process from the next. In just-in-time, the aim is
to couple all the processes together to minimise timescales and inventory.
Where kanban systems operate, the demand is smoothed to give similar
requirements every day, consequently there is not the need for significant
safety stock. Safety stock will be required where demand exceeds the flexi-
bility of the manufacturing process.
18 – an answer
Lean JIT throughput time is small; therefore, there is not the need for detailed
planning far ahead. This gives flexibility. By not having to forecast far ahead,
then the forecasting process is more accurate. Smaller batches mean continu-
ity of work and simpler planning; therefore, fewer problems and less cost.
They create regular tasks, providing businesses with great advantages over
their competition because the costs of storage, inter-process stocks and non-
value-added work are reduced.
19 – an answer
Table R19
Date Receipts Unit Issues Balance No left From Average

cost 14/3 10/6 cost
14/3 10 11 10 10 11
12/4 3 7 7 11
22/5 2 5 5 5 11
10/6 5 15 10 5 5 13
30/6 3 7 2 5 13
17/7 2 5 0 13
So, value of issue of the two items was:

a) FIFO costing = 22; Average value = 26; LIFO costing = 30
b) Highest stock value lines independent of usage rate. Reduce supply. See
Chapter 5.
20 – an answer
To absorb the different rates of supply and usage in two processes. The
calculation of safety stocks normally only takes into account the variations
in demand. The variation in supply is avoidable and should be controlled
properly so that it is insignificant. The variation in demand is caused by
the ‘customer’ and is therefore to be coped with and negotiated. The better
the forecast of demand, the smaller the safety stock will have to be. See
Chapter 12.
21 – an answer
The calculation of safety stock depends on:
• The level of service being offered (acceptable risk of running out).
• The predictability of the demand (MAD as a measure of forecast
accuracy).
• The replenishment lead time (but only as the square root of lead time).
See Chapter 7.
22 – an answer
a)
Table R22A
Item Wk 1 Wk 2 Wk 3 Wk 4 Wk 5 Average
D523 60 30 20 40 50 40
K488 30 15 0 65 0 22
Errors from average MAD Pmad
D523 20 10 20 0 10 12 30%
D488 8 7 22 43 22 20.4 93%
b) Lead time two weeks CSF 90% 1.6

CSF 95% 2.06
Table R22B
MAD SS 90% SS 95% Increase Extra cost % of turnover
D523 12 27.15 34.96 7.81 £3.90 20%

K488 20.4 46.16 59.43 13.27 £6.64 60%
£10.54
The additional cost is £10.54.

23 – an answer
Apply Pareto’s Law, then organise the supply to save work (see also question 12).
Change the order pattern for the 128 A class parts to weekly and schedule in
to avoid paperwork and time. Order the B class every four weeks as they fall
below the review level, and order the 896 C class items every eight weeks to
reduce the workload.Then work out the better safety stocks and flexibility of
supply for the A class. It is important to measure customer service level and
inventory levels while the change is happening, and also to make sure that the
number of deliveries is reducing significantly. See Chapter 3.
24 – an answer
The re-ordering system creates replenishment orders, which open the supply.
However, demand may not be satisfied because:
• From the theory: The review level is calculated assuming a customer
service factor which gives a small risk of stock-outs.
• In practice: Supply may be later than expected, damaged or poor quality
and the safety stock is insufficient to cope with the disruption.
25 – an answer
To set the safety stock requires:
Variability (MAD or standard deviation)
Availability policy
Supply lead time.
In addition, to set the review level and average stock level requires:
Delivery frequency or quantity.
Underlying the forecast and MAD calculations is therefore an assessment of
demand pattern using historical information backed up by understanding of
how future demand may vary from this.
26 – an answer
Average demand = 65 Safety stock = 16.67 MAD = 8.33
a) Review level = 81.67
b) Average stockholding = 66.67
27 – an answer
The re-order level is used to trigger new replenishment orders when the stock
+ supply orders outstanding become low. If the delivery frequency is
shorter than the lead time, then the stock is always low, delivery quantities
are small and there are always supply orders outstanding. Hence, for long
lead time A class items, delivery is frequent and stock is always less than the
review level.
28 – an answer
A class – order in daily – if possible on kanban.
C class – order in weekly.
See Chapter 10.
29 – an answer
Average = 24 MAD = 8
Customer service factor is 1.6
Safety stock = 1.6 × 8 × √(9) = 38.4
Review level = 24 × 9 + 38.4 = 254.4
b) The demand (rainfall) is not seasonal.
The product is not short life cycle (e.g. fashion).
The demand is a normal distribution (which is likely in this case).
Supplier lead time is correct and delivery is on time.
Competition remains the same.
Retailers are not doing promotions.
30 – an answer
a) Work out with the suppliers:
Benefits to supplier as well as customer (continuous demand, earlier
payment, more consistent demand); delivery arrangements (transport,
packaging); despatch and receiving reorganisation; information process-
ing (schedules and invoicing).
b) Reduction in inventory:
Average stock = half delivery quantity + safety stock. Assuming the
safety stock remains the same, delivery quantity goes from four weeks’
worth to one week’s worth, so stock saving is from an average of two
weeks to half a week.
At £1 million per year, 1.5 weeks is £28,846.
c) Benefits
Lower inventory investment – better cash flow
Less stores space required
Easier movement within stores
Improved co-ordination with suppliers (efficiency)
More competitive so growth and extra employment in the long term.
Disadvantages
More receiving document processing
Greater need for inventory accuracy
More frequent (but smaller deliveries) – so this may not be a disadvantage.
31 – an answer
The delivery quantity primarily depends on how frequently the supplier is
prepared to deliver without significant increase in cost. It will depend upon:
• The number and variety of other items bought from the same supplier.
• The type of items (e.g. handleable, slow moving).
• The quantity in which they are to be consumed.
• The turnover value of the item.
32 – an answer
Better forecasting.
Treating demand as dependent and avoiding safety stock effects.
Measuring end user usage; structuring supply source.
Integrating systems.
‘Square root of number of warehouses’ effect.
33 – an answer
Target stock level system: Variable order quantities; fixed time interval
between deliveries. Ideal where supply of many B and C class items are sourced
from a single supplier. Organise purchasing to look at suppliers in rotation.
Review level system: Interval between ordering is variable; order quantities
are fixed by an ordering rule. Normal way to trigger orders. Good for A class
items and slow-moving and one product suppliers.
See Chapter 8.
34 – an answer
a) Extrinsic forecasts to give causes of demand; market analysis and surveys;
historical techniques:
Moving average: OK, as demand is random and varying much
Exponential smoothing: Better, since there could be a slight trend
Double exponential: Will be over-reactive until a trend is seen
Base series: Not appropriate as seasonality not established
Regression: No advantage over other methods
More sophisticated techniques: Insufficient data for these
See Chapter 13.
b) Exponential smoothing with α = 0.2:
Table R34
Week Demand Forecast
10.00
1 12 10.40
2 14 11.12
3 8 10.50
4 16 11.60
5 13 11.88
6 18 13.10
7 ?
Forecast for week 7 is 13.1.

35 – an answer
MRP requires:
• Master schedule 95% accurate
• Bills of materials 98% accurate
• Inventory records 95% accurate
and
• Lead times (for suppliers and any manufacturing processes)
• Batch size rules
• Manufacturing routings (where appropriate).
36 – an answer
The system might recommend placing of additional supply orders for week 3
for module A and the batteries, assuming that the lead times allow. However,
this would create extra inventory for the rest of the schedule.
Table R36A
X Demand 1,000 1,000 1,000 1,000 1,000

Y Demand 500 500 500
Z Demand 100 100 100 100 100 100 100 100 100 100
Component Assembly requirements (offset by lead times)
A X and Y 1,500 0 1,000 500 1,000 0 1,500 0 1,000
Stock 1,500 0 0 –1,000 1,500 500 500 –1,000 –1,000 –2,000 1,000
Supply 3,000 3,000
New supply 1,000 3,000
Projected stock 0 0 0 2,500 1,500 1,500 3,000 3,000 2,000 5,000
B Z 50 100 100 100 100 100 100 100 100 100
Stock 1,500 1,450 1,350 1,250 4,150 4,050 3,950 3,850 3,750 3,650 6,550
Supply 3,000 3,000
Batteries 2X+2Y+Z 3,050 100 2,100 1,100 2,100 100 3,100 100 2,100 100
Stock 3,500 450 350 –1,750 7,150 5,050 4,950 1,850 1,750 –350 9,550
Supply 10,000 10,000
New supply 2,000
Projected stock 450 350 250 9,150 7,050 6,950 3,850 3,750 1,650 11,550
A better solution would be to bring the week 4 deliveries forward. A better option would be to try to reschedule the battery
supply to 1,400 per week and module A to 650 per week. An alternative in the short term is to extend the lead time to the
customers, as is illustrated in answer Table R36B.
Table R36B Working round supply issues
Week 1 2 3 4 5 6 7 8 9 10
X Demand 1,000 1,000 1,000 1,000
Y Demand 500 500 500
Z Demand 100 100 100 100 100 100 100 100 100 100
Component Assembly requirements (Offset by lead times)
A X&Y 1,500 0 0 1,500 1,000 0 500 0 1,000
Stock 1,500 0 0 0 1,500 500 500 0 0 –1,000 2,000
Supply 3,000 3,000
B Z 50 100 100 100 100 100 100 100 100 100
Stock 1,500 1,450 1,350 1,250 4,150 4,050 3,950 3,850 3,750 3,650 6,550
Supply 3,000 3,000
Batteries 2X+2Y+Z 3,050 100 100 3,100 2,100 100 1,100 100 2,100 100
Stock 3500 450 350 250 7,150 5,050 4,950 3,850 3,750 1,650 11,550
Supply 10,000 10,000
As excessive stock of module B is building up, changing the production plan to make more of watch Z and
less of X and Y would solve the supply problem, but leave a sales issue, as shown in answer Table R36C.
Table R36C Allocating the available production
Week 1 2 3 4 5 6 7 8 9 10
X Demand 500 700 700 700 500

Y Demand 400 500 500
Z Demand 100 300 300 300 300 300 300 300 300 300
Assembly requirements (Offset by lead times)
A X and Y 900 0 700 500 700 0 1,200 0 500

Stock 1,500 600 600 –100 2,400 1,700 1,700 500 500 0 3,000
Supply 3000 3,000
B Z 50 100 100 100 100 100 100 100 100 100
Stock 1,500 1,450 1,350 1,250 4,150 4,050 3,950 3,850 3,750 3,650 6,550
Supply 3,000 3,000
Batteries 2X+2Y+Z 1,850 100 1,500 1,100 1,500 100 2,500 100 1,100 100
Stock 3,500 1,650 1,550 50 8,950 7,450 7,350 4,850 4,750 3,650 13,550
Supply 10,000 10,000
37 – an answer
a)
Table R37A
Week
Stock 1 2 3 4 5 6 7 8 9 10
Sales 210 300 140 65 200 260 150 100 190 220
FG 650 440 140 0 0 0 0 0 0 0 0
Order receipt 65 200 260 150 100 190 220
Order release 0 65 200 260 150 100 190 220
Semi-finished 220 220 155 0 0 0 0 0 0
Order receipt 45 260 150 100 190 220
Order release 0 45 260 150 100 190 220 0
Raw material 500 500 455 195 45 0 0 0 0
Purchase receipt 55 190 220
Purchase ordering 0 0 55 190 220
b)
Table R37B
Week
Stock 1 2 3 4 5 6 7 8 9 10
Sales 210 300 140 65 200 260 150 100 190 220
FG 650 440 140 0 435 235 475 325 225 35 315
Order receipt 500 500 500
Order release 0 500 0 500 0 0 0 500
Semi-finished 220 220 220 220 220 220 220 220 220 220 220
Order release 500 0 500 0 0 0 500 0 0 0
Raw material 500 0 0 0 0 0 0 0 0 0 0
Purchase receipt 500 500
Purchase ordering 500 500
38 – an answer
a) DRP takes the forecast from warehouse and aggregates it into the master
schedule. It therefore depends upon the warehouses to forecast in the first
place.
Fair shares can manufacture against a forecast and then allocate the
inventory against the warehouse requirement.
b) The logic is:
1 Any unfulfilled demand from the warehouses.
2 Forecast usage which will be unfulfilled within the supply lead time.
3 Top up warehouse safety stock.
4 Allocate to central stock safety level if appropriate.
5 Distribute remainder in proportion to the forecast warehouse usage rates.
6 Transfer excesses to other warehouses if appropriate.
39 – an answer
Table R39
Week
Stock 1 2 3 4 5 6 7
London sales 150 320 370 190 150 320 379

Current stock 600 450 130 360 170 20 300 521
Order release 600 0 0 600 600 0 0
Manchester sales 70 70 70 70 70 70 70
Current stock 230 160 90 270 200 130 60 240
Order receipt 250 250
Order release 250 0 0 0 250 0 0
Newcastle sales 90 150 150 90 90 150 150
Current stock 310 220 70 10 10 10 40 70
Order receipt 90 90 90 180 180
Order release 90 90 90 90 180 180
Internal sales 30 30 30 30 30 30 30
Total 970 120 120 720 1060 210 30
Current stock 1650 680 560 440 720 660 450 420
Order receipt 1000 1000
Order release 1000 1000 0 0
Manufacturing orders required in weeks 1 and 2.

40 – an answer
Advantages:
• Retail stores satisfy customer demand and therefore stock is available
immediately to meet demand.
• Any distribution problems do not affect availability.
• No central stock is necessary.
• Slow distribution method can be used to the retail stores.
Disadvantages:
• More stock is required because there are more stock points.
• Any change of products requires emptying the whole supply chain.
• Redistribution of excess or supply to meet an individual shortage is
more difficult.
41 – an answer
Table R41
Item no Unit cost Annual Turnover % turnover Cumulative

usage % turnover
2 125 100 12,500 62.5% 62.5%

4 50 70 3,500 17.5% 80.0%
9 7 300 2,100 10.5% 90.5%
3 85 10 850 4.3% 94.8%
6 26 20 520 2.6% 97.4%
Item no Unit cost Annual Turnover % turnover Cumulative

usage % turnover
1 150 2 300 1.5% 98.9%

10 6 20 120 0.6% 99.5%
7 20 3 60 0.3% 99.8%
5 40 1 40 0.2% 100.0%
8 10 1 10 0.1% 100.0%
Total 527 2,000 527 20,000
a) A class items 2
B class items 4, 9
C class items 3, 6, 1, 10, 7, 5, 8
b) ABC determines delivery quantity and reduces purchase workload (see
Chapter 3).
c) Discussion of optimising profits, focus for supplier co-operation, cost
saving and lead time reduction.
42 – an answer
a) Independent demand:
99.2% customer service level requires CSF of 3
Average demand = 12, so delivery quantity = 12
MAD = 4.4, so safety stock = 3 × 4.4 × √(1) = 13.2
Average stock = half delivery quantity + safety stock = 6 + 13.2
Average stock = 19.2
b) MRP receives the week before and despatches the next week, so half a week’s
inventory results. Assuming no safety stock is required, stock would be:
Table R42
Demand 10 20 5 10 15 12.0
MRP
average stock 5 10 2.5 5 7.5 6.0
JIT
Daily demand 2 4 1 2 3
One day’s stock 2 4 1 2 3 2.4
Average stock = 6
c) Kanban supply: Receive and despatch within one day, so average
stock = 2.4, as shown in the table, assuming that the demand is known –
as with (b).
43 – an answer
Stock level depends mainly on supply delivery quantities, customer service
and demand pattern. Lead time variation is only as the square root. However,
delivery performance may be lead time dependent also.
Average stock = half delivery quantity + customer service factor × MAD ×
√lead time. So, changing the lead time has the least effect on stock levels.
44 – an answer
a) Fair shares logic:
Back orders.
Demand in lead time cover safety stock.
Fair share remainder reallocation.
Dangley: 19; Brassby: 20; Carbock: 16.
Table R44
Current stock 27
Delivery 55
Total 82
Priority Stock left
1 Overdues 2 80
2 Lead time use 30 50
3 Safety stock 30 20
4 Demand next week 40
Fair share 20/40 50%
Demand next week Fair share Total share Balance required
Dangley 10 6 26 19
Brassby 8 5 20 20
Carbock 12 9 36 16
55
45 – an answer
The answer should include the following concepts:
• Identify the historical usage data for each product.

• Use historical forecasting methods (exponential, DES or base series if
seasonal).
• Base forecasts on the historical information.
• Gain market information from sales and marketing in case future
demand is not a continuation of history.
Additional points can be:
• Consumer data from supply chain (end of supply chain).

• Understand promotions.
• Concentrate on A class.
46 – an answer
For company i).
Risk: Small company so risky, also depends on family maintaining interest in
the products. Supplier could dominate the small company, so may think it is
good idea; however, sales presence is going to be limited.
Merit: Dynamic and dedicated to the products, and expanding.
For company ii).
Risk: The distance to the market could be a problem, and also an expense.
There may be a different cultural approach, which will make it more compli-
cated. Sales forecasting could be poor – particularly capacity balancing if the
customer has too much clout. Could be a minor product supplier, so potential
not realised.
Merit: Worldwide sales presence – especially if branded products. Large
opportunities worldwide (China is probably an expanding market). Looks a
good match as long as management get on well together.
47 – an answer
Table R47
Week (t) Actual demand (thousands) (A) Forecast

(Ft = α At–1 + (1 – α) Ft–1)
1 55.3 56.0
2 54.5 55.8
3 57.8 55.4
4 55.0 56.1
5 57.2 55.8
6 61.2 56.2
7 55.6 57.7
Prediction for week 8 is 57.7.

Index
9-Box model 223–6 delivery frequency 122–4; quantities

140–55
ABC analysis 36–50, 223–4 deviation measurement 105
accurate records 38, 205, 219 distribution requirements planning
added value 32, 134–9, 140, 218 230–3
aggregate stock valuation 77 double exponential smoothing 189–94
allocation of stock 212–14, 226, 230–5
assessment of demand 226 economic order quantity 150–1
availability measurement of 19–21; electronic data interchange 26
of stock 101–3 enterprise resource planning 240–1
average value 101–3 excess and obsolete items 75,
110–12, 162
backward scheduling 203, 233 expediting 74, 82, 117
bar coding 163 exponentially weighted average 173–5;
base series 195–7 smoothing 175–7
batch 201, 205; sizes 215–16
Baysian forecasting 198–9 fair shares 234–5
bias 162–3 first in, first out (FIFO) 75–8
big data 241 forecasting demand 156–68; exponential
bill of material (BOM) 205, 208–10 188–9; focus 179–93; historical 169–87;
inaccuracy 162–3; methods for
collaboration 240 improving 166–9; seasonal 194–200;
company strategy 13, 67 techniques 169–70; tools 188–93;
consignment stock 109, 227–30 tracking 181–7
consuming forecast demand 28 Forrester effect 220–1
contributions of history 174 Fourier analysis 72, 199–200
customer feedback 28; relations 16–17, 85;
service 16–32; service factors 106–7; Gaussian 100
support 162
Cusum 182 Holt’s method 190, 200
data aggregation 167 independent demand 201–4

delivery promising 213–15; quantities 144 information accuracy 219; structures
demand driven MRP 221; history 99, 102, 240–3
114, 166, 170–1; management 24–8; initialising the forecast 178
patterns 101–3 inventory control objectives 4–5; forecast
dependent demand 201–16 types 158; investment 44, 226;
depreciation 8, 79 management 1–15, 217–37; valuation
despatch information 25–7 75–9
Index 277
just-in-time (JIT) 88–9 product life cycle 160–2, 170, 188
pricing methods 138–9
Kaizen 4, 69, 85 profit through inventory management
Kanban 88–9, 96 7–9
key performance indicators (KPI) 5, projected available 206–14
66–7 pull system 55, 86–8
Kraljic’s matrix 128–9 purchase commitment 69–71; order
57, 69, 94, 143, 154, 203–4
lag 181, 187; in exponentially weighted purchasing see procurement
average 190 push system 88
last in, first out (LIFO) 76
layout and location of stores 12 quality management 68, 92–3;
lead time 117–21 of suppliers 135–6
lean focus 84–6; implementation 80–97; queue 69, 86
principles 84–9; philosophy 80–4;
supply chain basis 217–19 record accuracy 162–8
logistics 6, 83, 85, 93, 217–18 recycling 236
lot for lot 215–16 reducing stockholding 51–61
regression analysis 195, 198, 201
management and control 62–79 reordering 113–19
manufacturing lead time 82, 120 repetitive processes 86–7; supply 141
market change 10–11, 168–9 replacement value 77
master planning 210; scheduling replenishment 42, 77, 122–7, 142, 154;
210–16 techniques 226–7
material requirements planning (MRP) reverse logistics 236–7
203–9 review levels 113–27
mean absolute deviation (MAD) 104–6, risk measurement 106–8
115, 178, 180
minimum stock levels 73; order constraints safety stock 98–112; evaluation of
152; order quantities 152 106–12
moving average 171–3 scheduling supply 148–50
seasonal sales forecasting 194–200
non-movers 55 setting stock targets 45–8
non-stock items 20, 41–2, 203 shaping inventory 33–50
normal demand patterns 101–3 single sourcing 132–3
skills and systems 71–4
obsolescence 5, 11, 58, 79, 88, 112 slow moving items 12, 56, 110–12,
order processing 13, 17, 24–9, 118; types 200
142; quantities 125, 143–4, 146, 150, smoothing constant 175–93
154, 215–16 standard cost 76; deviation 103–4
ordering 13, 49, 56–7, 113–19 stock availability 101–3; cover 43–7; levels
overheads 76, 95, 243 113–27; reduction 51–61
stockkeeping unit (SKU) 77, 226
Pareto 33–50; delivery quantities stores layout 12–13
145–6 supplier lead-time 118–19
perpetual inventory 38, 163 supply chain 128–35, 140–53; inventory
planned inventories 12 management 217–37; partnerships
planning the structure of supply 133–5; structure 222–3
209–10 systems and control 13–14
Poisson’s Law 101, 110
Pokayoke 71, 85 target stock levels (TSL) 124–7
pricing method 138–9 total quality management (TQM)
procurement 128–39 92–3
278 Index
tracking signal 181–5 variance 76, 105, 109, 179
training 66, 71–3 vendor appraisal 135–7; managed
transport algorithm 234–5 inventory (VMI) 49, 153, 227–30
Trigg’s tracking system
183–5 weighted averages 159, 173, 182–3
turnover of stock 43–5 work in process (WIP) 82, 205, 222
unit cost of stock lines 75–7 zero inventories 66, 241

Webpdf

Uploaded by

Copyright:

Available Formats

Webpdf

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Webpdf

Uploaded by

Copyright:

Available Formats

Best Practice in Inventory Management

Best Practice in Inventory Management, Third Edition offers a simple, entirely

ISBN: 978-1-138-29442-4 (hbk)

Typeset in Times New Roman

Visit the companion website www.routledge.com/cw/Wild

1 The basis of inventory control 1

4 Practical methods for reducing

8 Setting the right stock levels 113

10 Delivery quantities 140

11 Forecasting demand 156

12 Historical forecasting techniques 169

13 Improved forecasting methods 188

14 Dependent demand 201

15 Supply chain inventory management 217

16 Meeting the challenges 238

1.1 Which direction? 2

• students of supply chains, operations management or even purchasing;

Many people do not understand that professional expertise is required to control

• Why we need professional inventory management.

Everyone is an inventory controller, at home and at work. We all keep food,

Inventory management is a well-developed science, not simple

• Conflicts which beset inventory controllers.

1.A The role of inventory management

Figure 1.1 Which direction?

1 To cover up the mismatch between supply and demand processes.

Inventory control keeps balancing conﬂicting requirements.

Purchasing consider that inventory is a buffer enabling goods to be purchased

Reliability and consistency are the keys to effective inventory management.

Quality management often causes delays in processing materials and increases

Do not confuse high stock with good availability.

Forecast Sales ⇒ Determine Appropriate Inventory Supply ⇒ Replenish Inventory

1.B Objectives for inventory control

• the service that it intends to provide to customers;

Inventory management is simple:

The assets include both:

and the effect is clearly illustrated by the following example.

1.C.1 An example of the effect of stock on proﬁtability

Storage costs and inventory consume cash.

Cost element Cost (£ p.a.)

Stores: extra area rent, heating, etc. 500

Figure 1.2 Cost element.

Hold a minimum of the right items to be proﬁtable.

1.D Reasons for the current stock

Your current stock shows what is going wrong.

1.D.1 Supply quantity

Frequent, small deliveries keep stock low.

1.D.2 Safety stock

Let the source of the risk hold the inventory.

1.D.3 Market change

Detect trends and react automatically.

A sudden reduction in demand can only be anticipated through understanding

1 The marketing department puts items on promotion without informing the

1.D.5 Poorly deﬁned responsibility

1.D.6 Planned inventories

1.D.7 Layout and location of stores

1.D.8 Company strategy

1.D.9 Systems and control

• Enter issue from stock on system (any delay here?).

For efficiency, use real-time procedures.