FLEXE Dynamic Warehousing Strategies PDF
FLEXE Dynamic Warehousing Strategies PDF
FLEXE Dynamic Warehousing Strategies PDF
T
DYNAMIC WAREHOUSING
STRATEGIES
On‐demand warehousing for ecommerce
Karl Siebrecht
CEO, FLEXE. [email protected]
Amitabh Sinha
Associate Professor, Ross School of Business, University of Michigan.
[email protected]
Stephen Johanson
President, Starboard Solutions Corp. [email protected]
Vivek Rajeevan
Student Consultant, MBA Class of 2017, Ross School of Business, University of
Michigan. [email protected]
Tauber Institute White Paper, May 2016.
A webinar hosted by JOC accompanying this white paper is available
on‐demand at www.joc.com/webcasts.
Executive Summary
Dynamic warehousing is a warehousing strategy consisting of purchasing warehousing services
in a pay‐per‐use model, on an electronic marketplace.1 This can be the entire warehousing
strategy for a company, or it may be an activity that supplements an existing network built on
long‐term contracts. In either case, it allows the retailer to quickly adapt to variable demand
and cost conditions.
Retailers only pay per‐unit costs for the services they use, thus avoiding capital expense: this is
particularly useful in ecommerce, where retailers typically face high demand uncertainty while
also being capital‐constrained. Additionally, dynamic warehousing allows ecommerce retailers
to rent small units of capacity in many parts of the country, enabling quick delivery to a large
fraction of customers.
This white paper introduces the concept of dynamic warehousing, and demonstrates its value
via a case study.
Warehousing today
Warehousing is integral in retail operations, serving to distribute goods efficiently to the
locations where demand exists. Before the advent of ecommerce, many retailers used
warehouses as intermediate storage points (distribution centers or DCs) to supply their retail
stores. In this traditional or brick‐and‐mortar retail model, actual consumer demand is fulfilled
from retail stores. Therefore, a retailer could implement a DC network consisting of a small
number of DCs, each serving a “region” comprising many states, with replenishment lead times
of a few days.
According to CSCMP’s Annual State of Logistics Report2, total logistics expense in the US in 2014
was $1.45 trillion, equating to approximately 8.3% of GDP. Of this, $900 billion was
transportation costs. Although warehousing alone accounts for only 10% of total logistics cost
at $143 million, warehousing has a significant effect on transportation costs: a more extensive
network reduces outbound shipping costs (which is generally more expensive per unit), despite
some increase in inbound transportation costs.
1
This is also called on‐demand warehousing, although in this report we will exclusively use the
term “dynamic warehousing”.
2
CSCMP’s Annual State of Logistics Report, Council of Supply Chain Management Professionals,
Lombard, IL. 2015.
1
Ecommerce and its impact on warehousing
With the advent of ecommerce, however, it is becoming apparent that the traditional
warehousing model is insufficient to deal with market realities. Ecommerce is growing very fast
within retail: total US ecommerce sales in 2015 clocked in at $340 billion, comprising 7.5% of
total retail, and growing at a 14.7% annual rate3.
Ecommerce creates significant challenges in terms of customer expectations, compared to
traditional retail. Much of this is due to Amazon.com, which is the market leader in ecommerce:
by building an extensive distribution network along with delivering excellent customer service,
it has set up expectations in ecommerce that apply to all retailers.
Traditional retail (pre‐internet) Ecommerce
Outbound shipping is in larger quantities to Outbound shipping is in extremely small
retail stores. quantities to individual customers.
Outbound shipping time can be a few days, Outbound shipping time is critical, with many
because it is only replenishing store customers expecting items within 2 days or
inventory. less.
Demand variability is usually small, Demand can be highly variable, influenced by
influenced only by mass media with slower social media and faster news cycles in
news cycles. internet media.
Ecommerce is changing the nature of the warehousing industry as well. Ecommerce retailers
keep more inventory in warehouses, because of the lack of retail stores. As a result, demand for
warehousing space is growing, as is the need for an efficient warehousing and distribution
strategy. To quote the CSCMP Annual State of Logistics Report:
“Growing inventory levels filled all available capacity, with the national vacancy rate dropping
2.7% to 7%. Same‐day delivery models and expanding e‐commerce are reshaping warehousing
needs and expanding the penetration of warehouse management software to handle some or
many tasks that are well beyond the traditional warehouse offerings of the past. New operating
strategies and warehouse location requirements are changing the face of this sector.”
In this white paper, we will explore one such “new operating strategy” and its associated
“warehouse location requirements”: the idea of dynamic warehousing.
Need for alternative warehousing solutions
Most retailers have relatively small ecommerce volumes. The focus of this white paper is on
new or emerging ecommerce retailers, where volumes are even more likely to be very small.
3
Quarterly Retail E‐commerce Sales, 4th Quarter 2015, US Department of Commerce.
2
With small volumes, dealing with the challenges of ecommerce distribution listed above is
extremely difficult. Before dynamic warehousing, the following were the options faced by such
a retailer:
Startup – Drop ship: If the retailer owns its own manufacturing/assembly facility,
initially it may ship directly from that facility. Many small ecommerce retailers start this
way.
Self‐owned network: If the retailer operates its own network of warehouses, it is
unlikely to have the scale and financial resources to build a very extensive network. As a
result, the average distance to the customer is high, resulting in high shipping costs and
longer delivery times.
Network outsourced to 3PL: Although this offers a little more flexibility compared to a
self‐owned network, most 3PLs demand commitments of 1‐3 years. This effectively
locks in the retailer into a fixed structure for a period of a few years.
Distribution outsourced completely: Amazon, for instance, offers a service called
“Fulfillment by Amazon” (FBA), wherein Amazon distributes the retailer’s products
through its network. Although this can provide excellent service times to customers, the
costs can be high and many retailers are wary of handing over a core part of the
business to their main competitor.
Dynamic warehousing: what is it?
The main idea behind dynamic warehousing is that the shipper has access to a large network of
warehouses, and can activate warehousing services from bulk pallet handling to fulfillment in
small to large volumes and for relatively short amounts of time “on the fly”. For example, a
small ecommerce retailer may decide to create half‐a‐dozen different distribution points, for as
little as 50 pallets at each warehouse, with little to no fixed time commitment. The warehouse
provider would use its own labor and equipment to perform standard and optional services
such as receiving, shipping, case pick, item pick, packing, etc., and would charge the retailer on
a per‐unit basis.
Dynamic warehousing is “dynamic” in the sense that the retailer can change the configuration
frequently: based on demand conditions, warehouse space could be deployed at different
locations, for different volumes, in a dynamic fashion.
In such a system, the retailer incurs no upfront fixed costs, and gains significant flexibility. Of
course, the unit cost charged by the warehouse provider may be higher or lower than what
would be incurred by the retailer if it operated its own high‐volume high‐utilization warehouse.
But this is the benefit of dynamic warehousing: the retailer gains flexibility and avoids capital
expense, even if sometimes the per‐unit cost is higher.
3
To make this happen, one needs a two‐sided marketplace where warehouse providers can list
their space, and retailers find space. Additionally, because of the significant role of technology
in managing inventory, inbound and outbound shipments, and orders in ecommerce, the
marketplace would need to offer an Order Management System and an integrated Warehouse
Management System, that can link the retailer’s systems with the warehouse providers’.
Flexe.com4 is one such company operating in the United States that provides this service.
Compared to a traditional third party logistics provider, dynamic warehousing offers the ability
to rent significantly smaller spaces for significantly smaller amounts of time. This is analogous
to what is already common in many other industries: for example, in trucking, a retailer can (i)
operate its own fleet of vehicles; (ii) enter into long‐term contracts with large trucking
companies, or (iii) move loads by contracting with one truck at a time through a freight
exchange. Moving from (i) to (iii), we see the capital expense decreasing, unit costs increasing,
and flexibility increasing: these are the same tradeoffs in warehousing as one moves from
operating one’s own network, to contracting with 3PLs, to dynamic warehousing.
Given these tradeoffs, it should be clear that dynamic warehousing is not beneficial for
everyone. If a retailer already has large volumes with fairly predictable demand (e.g. Walmart),
then operating its own network will be the most cost‐efficient. Dynamic warehousing is best
suited for retailers facing small but highly variable demand, such as emerging ecommerce
retailers. For other mid‐size retailers, dynamic warehousing can play a role as a “filler” to
handle unexpected demand variability, complementing a primarily self‐operated and 3PL‐based
network.
Value for warehouse operators
Dynamic warehousing provides a positive value proposition for warehouse owners as well.
Building a warehouse can be expensive; any unused space has an opportunity cost. Even if a
warehouse owner/operator has long term contracts with retailers or 3PLs for much of its space,
any remaining space can be turned into a revenue‐generating asset by placing it on a dynamic
warehousing marketplace. Depending on the warehouse owner’s operating costs, opportunity
costs, and market dynamics, it can choose a price that may be more, or less, than the rates it
charges its existing clients.
4
www.flexe.com
4
Case study: Expanding ecommerce retailer facing variable demand
To illustrate the economics of dynamic warehousing and compare it to a traditional operation,
we consider a simple case study of an ecommerce retailer with a single plant in Ontario,
California (equivalently, the retailer imports all its goods into the port of Los Angeles). We will
then see how the retailer can use dynamic warehousing to expand its network. Some basic
characteristics of the retailer and its demand are as follows:
Utilization Level 95%
Total Annual order quantity 1 M
SKU count 500
Units per order 1.5
Plant location Ontario, California
Area 95,000 sq.ft.
Customer Demand Map Mirrors US population
Table 1: Basic characteristics of case study: base case
Test Case 1: One warehouse serving all customers in US
In its initial network, the retailer simply uses small package shipping to ship orders to its
customers, who are spread across the US. Using demand information, standard industry
transportation costs, and a fixed warehouse location near the plant, we can calculate the cost
of operating the network. For this, and all the network design in our study, we use Tactician, a
web‐based network optimization software5. The optimal network and resulting operating costs
are shown in Error! Reference source not found..6
Of the total cost of $2.25 M, about 80% is outbound transportation, from the warehouse to the
end customer. Inbound transportation rates are very low, but the distance is minimal. So, in
order to reduce transportation costs and in turn total costs and improve service level, it would
make sense to have multiple warehouses across the country. However, this also implies higher
warehousing costs as the fixed costs would increase as we add more warehouses to the
network.
Total Cost per order $ 11.89
Transportation cost per order $ 9.53
Order fulfilment within 2 days 12%
Total committed cost $ 2.25 M
5
Tactician is developed by Starboard Solutions Corp., and is available at
www.starboardcorp.com.
6
Only high‐level calculations are reported in this white paper. Complete details of calculations
are available upon request.
5
One Time Fixed cost $ 225 K
Three year locked in lease $ 2 M
Total annual cost $ 11.9 M
Figure 1: Network structure and operating characteristics for base case.
Also note that the existing network only delivers 12% of orders within 2 days. As the
ecommerce industry moves towards faster and faster delivery times (same‐day is increasingly
common), it is important for the ecommerce company that the customer service level is as high
as possible while maintaining reasonable costs.
Test case 2: 70% & 90% orders fulfilled within two days
We now consider the case where the goal is to deliver 70% of the orders within 2 days, at
minimal cost. Again using our demand and shipping information as described above, as well as
industry standard warehouse costs, we use Tactician to find the optimal network. The optimal
network has 3 warehouses, in CA, IL, and NC.
Total Cost per order $ 10.25
Transportation cost per order $ 7.76
Warehousing cost per order $ 2.49
Total annual cost $ 10.25 M
Figure 2: Network structure and operating costs to guarantee 70% delivery within 2 days.
6
Similarly, if we want 90% deliveries within two days we would require a total of 8 warehouses.
The network would look as below:
Total Cost per order $ 10.44
Transportation cost per order $ 7.48
Warehousing cost per order $ 2.96
Total annual cost $ 10.44 M
Figure 3: Network structure and operating costs to guarantee 90% service within 2 days.
This network has 8 warehouses7, reflecting the need to get closer to the customer. Many
ecommerce retailers are also offering same‐day shipping capability; to be able to serve 90% of
customers within 1 day, we would need a network of 16 or so warehouses.
For a small ecommerce retailer, the capital expense required to build out such a network would
be prohibitive. Locking into long‐term leases with 3PLs may also be risky, especially if volumes
are insufficient to gain economies of scale. This brings us to dynamic warehousing: suppose the
retailer was able to rent space, using current marketplace rates, for the same network?
$14,000,000
$12,000,000
$10,000,000
$8,000,000
$6,000,000
$4,000,000
$2,000,000
$-
Current Traditional (70%) Dynamic (70%) Traditional (90%) Dynamic (90%)
Startup Costs Committed Costs (Lease) Variable WH Costs
Figure 4 displays the costs of the three networks examined above, as well as the latter two
networks operated using dynamic warehousing. For dynamic warehousing, the network
structure is the same (i.e., warehouses are located in the same places as the corresponding
traditional network). However, there are no start‐up costs; instead, the dynamic network uses
7
Coincidentally, Walmart is also targeting free two‐day shipping using a network of 8 DCs,
supplemented by its retail stores. See http://fortune.com/2016/05/12/walmart‐shipping/.
7
marketplace costs for warehouse space which are only slightly higher per‐unit than in a
traditional network.
Of course, the relative advantage of the dynamic network depends on many factors: the actual
marketplace rates for warehouse space, demand variability, etc. Note that marketplace rates
for warehouse space may be greater than, or less than, standard rates with a 3PL. If the
warehouse provider has surplus space that will be otherwise unused, it may well offer below‐
market rates in the short term. On the other hand, in a high‐demand location, marketplace
rates are indeed likely to be higher than 3PL rates. We address the demand variability and its
impact on dynamic warehousing next.
Test Case 3: Growth scenarios
Let us now assess the value proposition of dynamic warehousing model in an uncertain
environment. We consider three growth scenarios: annual growth rates of 10%, 20%, and 30%
respectively. The resulting annual volumes are shown in Table 2.
Annual Order Volume (Forecast)
Annual Growth (Forecast) Growth rate (%) Year 1 Year 2 Year 3
Base 20% 1,056,000 1,267,200 1,520,640
Low 10% 1,056,000 1,161,600 1,277,760
High 30% 1,056,000 1,372,800 1,784,640
Table 2: Order Volumes under different growth scenarios.
To model the traditional system, we assume that the retailer is locking in capacity for 3 years
with sufficient capacity to meet peak demand. This means that the retailer would lock in
sufficient space to meet annual demand of 1,784,640, even though in year 1 utilization would
be much lower than that. This is conservative; the retailer may well be able to negotiate
capacity reservation with the 3PL that does not require so much unused space in Year 1.
Nevertheless, for our calculations, we assume that the full capacity is reserved up front.
High growth (30%) Moderate growth (20%) Low growth (10%)
Utilization: 95% Utilization: 75% Utilization: 58%
$12.00 $10.72 $12.00 $12.00
$11.04 $11.40
$10.00 $9.31 $9.32 $9.31
$10.00 $10.00
$8.00 $8.00 $8.00
$6.00 $6.00 $6.00
$4.00 $4.00 $4.00
$2.00 $2.00 $2.00
$‐ $‐
$‐
Traditional Dynamic Traditional Dynamic
Traditional Dynamic
(90%) (90%) (90%) (90%)
(90%) (90%)
Figure 5: Cost per order with variable demand growth.
8
Figure 5 shows the cost per order incurred as the actual growth rate varies, assuming the firm
installed capacity for the high growth scenario. In the leftmost graph, actual growth was indeed
the 30% that was expected, leading to cost per order of $10.72 and $9.31 for the traditional
and dynamic networks respectively. The next two graphs show the costs if the actual growth
rate was moderate (20%) and low (10%) respectively. In those cases, the cost per order with a
traditional network rise to $11.04 and $11.40 respectively, because utilization decreases. Cost
per order under the dynamic model, however, stays unchanged at $9.31 because of the pay‐
per‐use nature of the dynamic warehousing system.
Of course, the retailer may choose to invest assuming a low growth scenario, and not risk low
utilization. In that case, the opposite problem will arise: if actual growth is high, then the
retailer runs out of capacity, and risks either losing demand or having to pay other costs to
meet the unexpectedly high demand.
This ability to deal with uncertainty is one of the key benefits of dynamic warehousing.
Uncertainty and variability arise in many forms, as we will discuss shortly. It should also be
pointed out that a retailer may deploy a strategy that mixes traditional warehousing with
dynamic warehousing: operate some warehouses which it owns or contracts with 3PLs, and use
dynamic warehousing as a filler when needed. That may indeed provide an even better level of
costs and service levels.
Other pros and cons of dynamic warehousing
Dynamic warehousing offers the ability to hedge against other types of variability, in addition to
the volume variability analyzed in this white paper. For instance:
Regional variability: Demand could grow much faster in some regions than in others.
Dynamic warehousing provides the ability to increase warehousing capacity in regions
with high/fast‐growing demand, and decrease warehousing capacity in regions with
declining demand.
Cost variability: Operating costs could change in different ways across different regions.
For instance, tax policies in one state could make costs particularly attractive there,
while wage and rent inflation could make other regions unattractive. A dynamic
warehousing network can quickly adapt to such changes.
Supplier variability: As the ecommerce retailer grows, it may add new suppliers in other
parts of the country. Or, the supplier itself may add new facilities, or transportation
disruptions may make the flow of imports arrive through a different port. Again,
dynamic warehousing can quickly adapt to these changing conditions.
9
It should be mentioned that dynamic warehousing is not without its own risks. The single
biggest risk is that the retailer is exposed to market rates for warehousing space. That is, much
like Uber’s surge pricing, market conditions may cause warehousing rates to spike suddenly. A
retailer who has its own network of owned/operated warehouses will typically be in a better
position to manage its costs, while a retailer exposed to the market for warehousing services
via dynamic warehousing is exposed to this risk.
Another potential risk factor arises from the fact that orders are being fulfilled by a network of
unrelated warehouses contracted only through an electronic marketplace. As in any such
outsourcing situation, the retailer is subject to the risk that operating conditions at some
warehouses may not be optimal, leading to, for example, errors in order fulfillment or
misalignment with the retailers values and objectives. This risk can be mitigated by appropriate
contract structures and monitoring, both on the marketplace platform and via third parties.
We also note that dynamic warehousing does not present an either‐or decision with respect to
traditional warehousing. A firm may well find it even more advantageous to operate a hybrid
network, with some fraction of inventory and orders going through its self owned/operated
network and the remainder through a dynamic warehousing network. Such a system can offer
most of the advantages listed above while also lowering the risk exposure; however, it would
require some amount of capital expense.
Conclusion
Over the last few years, information and communication technologies have led to the
emergence of electronic marketplaces for a number of services and business activities. The
most famous of these is of course Uber, but these marketplaces exist for everything from
simple human tasks to outsourced manufacturing abroad. This phenomenon is perhaps most
accurately called “platform capitalism”,8 and the most significant enabler of it has been the rise
of cheap and high capacity information and communication technologies. Recent trends
suggest that platform capitalism is only going to grow, in both B2B and B2C settings, as
companies focus on their core competencies and take advantage of the flexibility and cost
benefits that come with a well‐functioning marketplace with multiple suppliers of all the
services a business needs to operate.9
Dynamic warehousing can therefore be seen as an instantiation of platform capitalism in the
warehousing services industry. As overall warehousing costs continue to rise, and rise at
8
See, for example, https://www.theguardian.com/technology/2015/jun/07/facebook‐uber‐
amazon‐platform‐economy and http://networkcultures.org/mycreativity/2014/10/16/never‐
mind‐the‐sharing‐economy‐heres‐platform‐capitalism/.
9
The Vanishing American Corporation by Gerald Davis is a recent book that strongly makes this
case.
10
dramatically different rates in different parts of the country,10 dynamic warehousing offers
emerging retailers the flexibility to compete in such an environment. Of course, reserving and
operating warehousing services is more complicated than hailing a ride on Uber, but that is
where software systems and technology can provide efficient solutions.
In summary, there is a strong case that not only does dynamic warehousing offer a cost‐
efficient way for small ecommerce retailers to offer high service levels in a flexible fashion, but
in fact it is likely to become a standard way of contracting for warehousing.
Acknowledgements
This white paper was supported by Flexe.com. Ishan Khandelwal’s initial research assistance is
gratefully acknowledged.
10
As reported in the Wall Street Journal on May 16, 2016, ecommerce has driven warehousing
rates overall to rise 9.9% in one year, with the San Francisco bay area seeing a 28.9% increase.
See http://www.wsj.com/articles/e‐commerce‐drives‐surge‐in‐u‐s‐warehouse‐rental‐prices‐
1463391930
11