A Roadmap For Scaling Self-Service Analytics 1.0

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

The Road Forward

A Roadmap for Scaling


Self-Service Analytics
Table of Contents
Change Isnt Coming. Its Here ........................................................3

The Modern Approach .....................................................................4

The Traditional Way ..........................................................................4

The Enabler is New Technology.......................................................5

The Barrier is Process......................................................................6

Applying Agile Principles to Analytics..............................................8

The Path Forward..............................................................................9

Fast Prototyping...................................................................................................................10

Clear Enablement Role for IT...............................................................................................10

Workflow ..............................................................................................................................10

Skilled Teams.......................................................................................................................10

Managing Change To Achieve An Analytic Culture ........................................................... 11

Conclusion............................................................................................................................ 11

Your Next Step ................................................................................. 12


Change Isnt Coming. Its Here.
Everyone in the world of data knows the signs. The amount and variety
of data makes the concept of one-stop data warehouses obsolete,
even as many companies still struggle to build their first data warehouse.
Business users, adapted to user-friendly consumer technologies, demand
the ability to work directly with their data. Technologies exist that allow
interactivity and manipulation of data in ways that were unimaginable
just years ago.

At first, this shift seems threatening, even out of control. It need not be.
In fact, it can reduce the crushing workload of dashboard requests so that
IT can focus on large and strategic issues. This elevates IT from the role
of dashboard factory to architect and steward of the companys assets.
And it frees business users from the slow, disheartning cycle of change
request and response. Self-service analytics can yield huge business
and employee dividends while protecting data assets and providing the
best source of truth out to the enterprise.

But even with new technologies that empower business users,


companies sometimes fail in their analytics strategies. New approaches
demand a new methodology. We look to proven agile development
and deployment methods that move as quickly as the changing
requirements. We look to a methodology that allows IT and business
to work together as partners. We look to a process that allows people to
exercise their natural creativity and curiosity. This is Tableau Drive: a
methodology that draws from agile methods and is informed by practical
experiences at the most analytically-minded companies in the world.

3
The Modern Approach
Facebook is a company that has achieved mass adoption of analytics.
At Facebook, employees try to inform every decision with data.
Business people are responsible for doing analysis. IT is responsible for
managing and securing data. Each team respects what the other brings
to the table. Both are advancing Facebooks ability to answer questions
and, in doing so, adding tremendous business value.

Namit RaiSurana, the Data Product Manager at Facebook, says there is


never a dependency on any of us to answer these questions, he said.
Users can discover for themselves what their data has to offer.
We are opening up Tableau to the entire company, he said.
Business intelligence dashboards are possible without having to
spend weeks programming.

This is all made possible by what business users dont see: the data
sources that are set up and managed by IT. This is a key concept: to
make the most of a self-service analytics strategy, you need highly
usable, easy to access data. The best analytics implementations are
user-created dashboards running on top of IT-managed infrastructure.

The Traditional Way


People within organizations have traditionally accessed data via static
reports from enterprise applications and business intelligence platforms
maintained by IT departments. These systems, predominantly designed
and built in the 1990s, are generally heavy, complex, inflexible and
expensive. As a result, business users are forced to depend on specialized
resources to operate, modify and maintain these systems. This creates
a divide between users seeking insight and technical specialists lacking
business context. This divide limits the usefulness of these legacy
systems. Because most business users lack the time and skills to bridge
the divide, they simply didnt use the analytics systems provided by their
companies.

As a result, many knowledge workers today rely on spreadsheets as their


primary analytical tool. Much of this is a failing of technology and much
of it is a failing of process.

4
The Enabler is New Technology
A new generation of business intelligence products has emerged to
support the inquisitive information worker in real-time. These products
have streamlined, direct-interaction interfaces that simplify shifting
perspective and provide immediate feedback to the user.
Without specialized training, information workers can engage in
interactive, iterative cycles of question-and-answer with data.

This technology has several key characteristics:

Easy to use. As the first generation of analytical systems proved,


tools that are hard to use will simply not be adopted by most knowledge
workers. Getting data into the hands of people who make decisions
requires a user experience that is not only easy to use, but also fun
and engaging.

Fast. Google engineers found that visits to a website will decrease if that
website loads even 250 ms more slowly than a competitors.
Speed matters in every context, research shows. Four out of five online
users will click away if a video stalls while loading.1 The bar has risen.

Powerful. Personal productivity software lets people get what they want
without going through an intermediary. People have experienced this with
word-processing software, web creation software, and recently video
production software. They want the same level of power when working
with data.

Visual. People are finding that working with data visually increases the
speed and quality of data insights. As data volumes continue to increase,
visual representations become even more important.

Business users know what questions to ask and how to interpret the
results. But business users may not knowor care aboutwhether the
system is scalable or secure. They begin to care when performance slows
down or sensitive information is compromised. Self-service analytics
works when the business users can answer business questions and IT
can ensure performance, security and data integrity.

http://www.nytimes.com/2012/03/01/technology/impatient-web-users-flee-slow-loading-sites.html

5
The Barrier is Process
Enabling people to analyze their own data has gone from a trend
to a norm. Most IT departments agree that self-service is a superior
approach, and are only too happy to get out of the report-writing
business. But IT does want to stay in control of critical aspects of the
system such as scalability, security and governance. And it should.
But traditional deployment models focused only on the role of IT and
assumed a role for the business only at the very beginning, in the
requirements phase. This effectively put IT in charge of any changes to
the requirements or evolution in the businessand cut the business out.

Most software deployment models, including those for business


intelligence, have been based on the waterfall or serial build
method of software development. Let us be clear that with the benefit
of time and well-known requirements, traditional serial-build or
waterfall software development can work well.

Requirements Plan Develop Test User Acceptance Production

Impeccably executed waterfall methodologies have ensured quality and


saved lives on complex civilian and military projects. As an example,
for over a decade the Onboard Shuttle project operated at Capability
Maturity Model Level 5 with 420,000 lines of code2 and an additional 1.7
million lines of Flight Software Application tools. Software controlled
every aspect of flight including liftoff, booster staging, return to launch site
(on abort), main engine cutoff, tank separation, on-orbit operations, entry,
energy management, approach, and landing. All this was achieved with a
waterfall methodology.

However, waterfall doesnt always work.

Bill Curtis, et all. The Capability Maturity Model: Guidelines for Improving the Software Process. 1994.

6
In 1997, Edward Yourdon wrote a book called Death March: The Complete
Software Developers Guide to Surviving Mission Impossible Projects,
which articulated the dread and emotional anguish associated with
meeting every fast moving business targets using traditional software
methodologies.

He noted that in the 1970s and 1980s, it was not uncommon for projects
to take 3-5 years from initiation to full-scale deployment. By the 1990s,
client-server and 4GL technologies allowed compressing substantial
projects into the annual budget cycle. By the late 1990s, 7-year projects
that had become 7-month projects were now hitting production in
7-weeks. Lighter methodologies that compressed the traditional cycle
were created and evangelized.

These so-called agile development approaches had many commonalities.


In the old model, developers and business negotiated a work agreements
and until code was complete and a build was released for
user acceptance, business agreed not to change requirements in
exchange for getting everything development agreed to do.

In the new models, large projects were broken down into faster, smaller
stages. Requirements were less formal and more flexible; and developers
were given discretion to change or improve them. Business, development
as well as documentation, testing and release teams were engaged
continually and collaboratively. Working builds were released on an
ongoing basis. Faster, shorter cycles allowed faster failures and faster
success but not all-or-nothing milestones that were often obsolete by
the time they were achieved.

7
Applying Agile Principles to Analytics
Agile approaches can largely be reduced to the shortening and softening
Plan of classic waterfall cycles. These cycles, or sprints, are then wrapped
closely around core subject matter expertise. The guiding principles of
Develop agile methods are:
Production Subject Matter
Expert People over processes and tools
(Requirments)

Working software over comprehensive documentation


Test
User
Acceptance
Collaboration over requirements gathering

Responding to change over following a plan

In an agile process, there is almost no benefit to exhaustively


documenting requirements. Requirements are much more flexible,
and the penalty to changing them is hours or days, not months.
This allows more feedback to come into the process earlier, resulting
in a much more effective result.

This is the idea behind Tableau Drive. Tableau Drive is a business


intelligence methodology that draws from agile methods and is informed
by the most analytically-minded companies in the world. With Tableau
Drive, business and IT jointly own the analytics platform. There is a
division of labor that plays to the interests and strengths of each.
And there is the opportunity to continuously evolve and grow, even at
the scale of an enterprise deployment.

This is the opposite of the state of business intelligence at many


organizations, where reports are returned after many rigid stages of
development and often after their period of usefulness.

Adopting Tableau Drive brings more than better analytics. It also supports
a faster and more flexible business. Think of how relationships changed
once written postal mail correspondence was replaced by phone calls.
How did business change when email replaced inter-departmental manila
envelope mail? Projects moved faster, and inefficienciesprinting,
finding envelopes, inter-office deliverywere eliminated, making it faster
and easier to iterate on ideas. It reduced a great deal of drudgery for office
workers and replaced it with the possibility to collaborate with co-workers
in less formal, more engaging, and ultimately more productive ways.
This is the promise of Tableau Drive: that a better analytics system can
lead to a fundamental change in the culture of an organization.

8
The Path Forward
Business users are developing the expectations, and the expertise,
to analyze data on their own. The waterfall development model,
which relies on hard upfront requirements and works at glacial speed,
has often failed: stretched-thin IT departments cant possibly build what
the business wants when the business can't possibly know all the
questions it will want to ask.

The Drive methodology takes the iterative process of creating analytics


and puts a structure around it to support an agile enterprise
deployment. In the Discovery stage, business and IT come together to
describe a vision for their process. Prototyping begins immediately,
with whatever data is available on hand, no matter how ugly it is.
The lessons learned from prototyping serve as inputs into the
foundation-building phase, when IT creates enterprise-wide data
sources, training programs are developed, and business users evangelize
the system using results from prototyping. If the foundation has been
built effectively, the organization is in a position to enable many new
groups and users.

Tableau Drive outlines a plan to enable self-service across the enterprise.


Its main tenets are:

Plan

Develop

Produc tion
Subject Matter
Expert

Test
User
Acceptance

Discovery Prototyping Foundation


Building

Scaling Out

9
Fast Prototyping
Business users have the ability to analyze data whenever and however
they can, starting immediately with whatever data is on hand.
Prototyping is no longer a rogue activity that gets thrown away once the
real project starts. It provides insight into how to create data sources
and how to train new business users. And, importantly, the results of the
prototyping are useful analytics products themselves.

Clear Enablement Role for IT


In Tableau Drive, IT owns the infrastructure that enables self-service
analytics. Production systems should be overseen by IT, which will
manage and provide data security, integrity, and the dissemination of
accurate information. IT will be responsible for defining production
data-sources, documenting data dictionaries, and acting as an integral
part of the analytics Center of Excellence. IT should be part of the
discussion as early as possible, ideally in the Discovery phase, and
should lead the discussion on topics such as governance.

Workflow
IT should create a sandbox that allows business analysts to query
and collaborate in small groups on their own and without supervision. IT
will delegate and supervise the process whereby reports (or workbooks)
are certified and promoted from sandbox to production.

Skilled Teams
IT will work as part of teams with the following skill-sets:

Understanding of the data

Understanding of the software

Understanding of the business

There is no right or wrong size for a team, but the test of its composition
will be the ability to analyze and make sense of data in real-time.
Sometimes consultants will be required to augment teams while business
users are ramping up.

10
Managing Change To Achieve An Analytic Culture
Fostering a culture of creativity where users are continually probing for clarity
is a mission that extends well beyond the edges of a software product.
Data driven decision-making requires programmatic support throughout
the organization. IT, as the ultimate owner of the analytics solution, has an
interest in and ability to foster intelligent usage of analytics in a way that
drives the business forward.

Ultimately, Tableau Drive puts IT in a strategic design role. IT is building the


car, the business is driving it. They must work together to get ahead quickly
and safely.

Tableau Drive, at its core, is an exercise in change management.


If your organization has serious impediments to changefor example,
a war between business departments and IT, or a fundamental aversion
to providing business users with datayou should plan to address those
issues early on in the process.

As part of the Drive process, teams should meet and agree on a mission
statement. Getting both business and IT to agree to a common mission
statement is an absolute requirement of creating an analytic culture.
Here is a suggestion that can be modified to fit the needs of
different organizations.

11
Drive Mission Statement Example
We are committed to running a smarter business. To do this we need to
empower people throughout the organization to ask and answer their own
questions. This will mean every single person can make better decisions,
every day.

The Drive Team will start this process of change by building the data structures
and reports to answer our pressing questions today. We will be agile in our
approach. This may involve changing data structures, building new data
structures, or performing new ETL functions to provide clean, usable data.

The team also has a mission to enable the organization to answer the pressing
questions of tomorrow. We will do this by providing a platform for the business
to explore, develop insight, and share insight with others. Using the platform,
from finding data sources to publishing and sharing content, will be as self-
serve as possible.

Drive provides a vision for what an analytically-enabled organization


looks like, and a road to get there. It requires openness and a willingness
to change.

Your Next Step


This document provides the vision for an analytics-enabled organization.
The Drive Manual is a comprehensive how-to document that outlines the
necessary steps for an enterprise deployment of analytics. It can act as
your guidebook. If your organization is ready to move forward, we
recommend that document as a next step.

12
About Tableau
Tableau Software helps people see and understand data. Offering a revolutionary new approach
to business intelligence, Tableau allows you to quickly connect, visualize, and share data with
a seamless experience from the PC to the iPad. Create and publish dashboards and share them
with colleagues, partners, or customersno programming skills required. See how Tableau can
help your organization by starting your free trial at tableau.com/trial.

Additional Resources
Start a Free Trial

Modern Approach to Enterprise Analytics

Related Whitepapers
Tableau Drive Manual

The Modern Approach to Business Intelligence

Tableau for the Enterprise: An IT Overview

Fostering a Data-Driven Culture by The Economist Intelligence Unit

Gartner positions Tableau as a leader in Business Intelligence

See All Whitepapers

Explore Other Resources


Product Demo

Training & Tutorials

Community & Support

Customer Stories

Solutions

Tableau and Tableau Software are trademarks of Tableau Software, Inc. All other company and
product names may be trademarks of the respective companies with which they are associated. 458780

You might also like