Professional Documents
Culture Documents
How to identify high-value opportunities for embedding analytics into your business processes
provided by
Successfully Automating Decisions with Business Analytics
Table of Contents
Introduction.................................................................................................................................... 1
The goal: evolve from “craft” to “industrial” analytics.......................... 2
Where to start: identify the opportunities......................................................3
Taking a Process View........................................................................................................... 5
“Big Picture” Frameworks................................................................................................. 6
Systematic Inventory............................................................................................................. 9
Setting Ambition....................................................................................................................11
Qualify the opportunities.................................................................................................13
DELTA-Readiness..................................................................................................................14
Motivation and Momentum............................................................................................14
Experimentation....................................................................................................................16
Summary........................................................................................................................................ 17
About nGenera......................................................................................................................... 18
About Intel.................................................................................................................................... 18
About SAS..................................................................................................................................... 18
i
Successfully Automating Decisions with Business Analytics
ii
Successfully Automating Decisions with Business Analytics
Introduction
Business analytics has been a rather ad hoc activity in most corporations, requiring
an interested decision-maker, a skilled analyst, and typically some unstructured activities
for pulling together the right data and constructing the right analytical model. There
is normally a high degree of interaction and iteration between analyst and decision-
maker as the model is refined and the results are understood. This approach can
serve individual decisions and initiatives (e.g., marketing campaigns) well, but it has
several shortcomings:
• It is time-consuming and labor-intensive, making it infeasible for decisions that
must be made often and quickly.
• The process steps and cycle time for ad hoc analytical decisions are
unpredictable, making it difficult to respond quickly to business change and
customer demand.
• Because the means of making the decision are typically not saved or embedded in
an ongoing business process, execution of the decision may be inconsistent, and the
next similar decision requires that the entire cycle be repeated.
The problem is that a process orientation is still not well established in many
corporations. They don’t have agreed-upon maps of major processes or owners
responsible for those processes. Even those that have mapped the activity and
information flows of their processes have yet to inventory the major decision points
and the potential for more analytical decision making.
So how do you determine which of your business processes would benefit the most
from efforts to embed analytics into them? Where would you likely get the biggest
return on your investment of time and money? The purpose of this paper is to
provide practical advice on how to identify the best opportunities for making your
processes more analytical.
1
Successfully Automating Decisions with Business Analytics
Business decision-making has been predominantly a “craft.” Each decision is its own
ad hoc effort, including decisions supported by extensive data and analysis. The
contemporary sequence runs like this: The company stores data from transaction
systems in data warehouses. A decision-maker defines the decision to be made and
communicates this to an analyst. The analyst reviews the data available, selects the data
wanted, extracts the data (sometimes with the help of IT), analyzes it with the help of
statistical software, and communicates the results to the decision-maker. The decision-
maker then cogitates, and the definition-extraction-analysis sequence may iterate, until
the decision-maker makes the decision. The approach is highly customized and often
skilled labor-intensive and time-consuming.
• Competitive pressures – for increased agility and a faster “insight to action” loop.
2
Successfully Automating Decisions with Business Analytics
Craft Industrial
Pattern Ad hoc, project-oriented Embedded in an ongoing process
Purpose One-time decision or event support Ongoing process performance
Benefit One-time Recurring
Higher up front, recurring to maintain
Investment Lower, one-time
the model
Time to
Brief Longer
Implement
Speed of Once implemented, fast or
Same as time to implement
Analysis instantaneous
Labor-intensive up front, analysis
Staff Labor-intensive
automated thereafter
Memory of
Can be saved for reuse, but often lost Maintained and improved upon
Analysis
Figure 1: From “Craft” to “Industrial” Analytics
There are two fundamental approaches to finding business opportunities for applying
analytics, and you should exercise both regularly:
• “Big picture” thinking – keeping tabs on what others in your industry are
doing, using business decomposition and analysis frameworks to assess where
performance can or must improve, and exploring your hunches about the
business, what makes it tick, and where the next breakthrough may await.
Most businesses start the opportunity finding process by surveying what’s happening
elsewhere in their industry. Figure 2 lists some of the most common applications of
analytics that you’re likely to find. Looking across your industry – and scrutinizing the
activities of your competitors – on a regular basis is essential. At a minimum, it keeps
you alert and informed about what you must do to keep pace not only with the
competition, but also with changing and rising customer expectations.
3
Successfully Automating Decisions with Business Analytics
• Something that you (or someone influential in the business) passionately believes
is important to customers. What needs or forms of value have customers
themselves not yet discovered or articulated?
There’s nothing wrong with the informed hunch or educated guess at the start of
opportunity finding, but you must verify your intuition by doing a small-scale test of
the idea and expanding if it seems to be working and making a difference.
4
Successfully Automating Decisions with Business Analytics
In thinking about how analytics might support your business processes, it’s important
first to ensure that you are thinking about your business activities from a process
standpoint. This means looking at the activities from an end-to-end perspective, with
the primary customer of the activities as the recipient of the process outputs. Taking
a process perspective allows organizations to identify all the possible ways in which
analytics might be used to create a better outcome for a process customer.
For example, Chad Jones, a telecom industry executive (Nextel, Sprint, and Grande
Telecommunications) has applied a process perspective to both customer and
employee relationships. At Nextel (which was acquired by Sprint), and later at the
mid-sized firm Grande, he created a “customer experience lifecycle” consisting of six
lifecycle stages (aware/learn, need/buy, install/use, pay bill, seek help, renew/upgrade).
For each stage, Jones and his colleagues identified a series of attributes and metrics.
They viewed each stage and attribute in the entire customer experience as being
potentially influential in determining customer loyalty. At Nextel between 2001 and
2005, this process-based approach led to a 40% reduction in customer turnover or
“churn,” while the subscriber base grew at over 100%. The focus on upsell and cross-
sell also substantially increased lifetime customer value.
Jones and his colleagues began to analyze employee attrition in 2003, and quickly
discovered that Nextel employees were most vulnerable to leaving within 45 days
of being hired. Nextel learned that new employees during this period often felt
ineffective because their computers weren’t installed and configured quickly, and
because supervisors didn’t provide enough attention and direction. When they
addressed some of the onboarding issues, turnover rapidly fell. Analysis also suggested
that signing up for the company’s 401K plan was critical to new employees staying for
more than a year. Managers intervened with employees slow to sign up, explaining
details such as the company contribution match, and turnover fell again. Just as with
customers, a process perspective was the key to better serving employees.
Most organizations already have some process orientation, be it Six Sigma, Lean, or
reengineering. The key is to identify which attributes in the process are associated
with satisfaction and value from the standpoint of the end customer. The analytical
work necessary to pinpoint the key attributes is not likely to be difficult once the
process perspective has been adopted.
5
Successfully Automating Decisions with Business Analytics
We recommend three tools for thinking broadly about your business, its performance
drivers, and its opportunities for differentiation.
The first is the Value Tree, a formal decomposition of business or shareholder value
most often used in decision alternative action analysis. The basic form of the tree goes
value-goals-objectives-criteria-attributes and then how specific decisions or scenarios
influence the attributes. Figure 3 has a simple example adjusted to depict major
business drivers and the processes whose performance most influences them. If you
have a comprehensive measurement system, you can perform statistical analysis on
your value tree and determine the correlations among business drivers.
New Markets
(e.g., geographic, vertical, segments)
Revenue
(Growth) New Adjacent Businesses
Share of Account/Wallet
Pricing/Margins
Procurement
Cost of Goods Sold
Supply Chain
Costs
(Productivity) Selling and Marketing
7
Successfully Automating Decisions with Business Analytics
The potential and therefore emphasis of business analytics varies by business type,
as outlined in Figure 6. In a value chain business, analytics may focus on problems of
supply/demand fluctuations, cost of assets, flexibility of operations, and interfaces with
others in the chain. In a value shop business, which incorporates new information to
solve new versions of problems, the most valuable analytics may be in support of
exploration and solution and decision design. In a value network business, analytics
may focus on understanding the network and its transaction patterns, the better to
attract quality customers and minimize cost. For a complete description of these
three business types, see “Casting off the chains: Value shops and value networks”
by Fjeldstad and Andersen (European Business Forum, summer 2003).
Note that a business of one type may have major processes more akin to a different
type. For example, a bank’s back-office operations may have a lot in common with
a value chain business, and its commercial loan department may function much like
a value shop, while the bank as a whole is a value network, enabling the exchange
of money. It’s useful to apply this business type analysis to major processes within a
corporation; however, you must keep in mind (and not work in opposition to) the
overall purpose of the business. For example, telecommunications companies that
run their customer service call centers too much like value chain businesses (pursuing
efficient utilization over customer service) do the overall business the disservice of
driving customers out of the value network.
Value type analysis can serve two important purposes in finding opportunities to
leverage business analytics. First, it can help you focus on the fundamental objectives of
the business and how analytics can serve them. Second, it can suggest where to look
outside your industry for fresh examples of analytical applications. When looking across
industries for inspiration, your best hunting grounds are likely to be other industries
within your value type (e.g., telecom companies learning from financial institutions).
These are companies that face analogous business problems but in different contexts,
and their analytical applications may be recognizable, relevant, and transferable.
8
Successfully Automating Decisions with Business Analytics
Systematic Inventory
Another way to find specific opportunities for applying analytics is by looking in detail at
your processes and how decisions are made within them. What kinds of processes lend
themselves to – or call for – business analytics? They tend to be processes that are:
• Information-intensive – analytics reveal the meaning of information.
You can also find specific opportunities for applying analytics by inventorying your key
business decisions (whether or not they are associated with well-defined business
processes) and asking how better information and analysis would make for faster and
better results. In general, look for:
• Complex decisions with lots of variables and steps within the decision process.
• Places where you want to optimize the whole process or activity (i.e., if
you simply decompose and optimize locally, you sub-optimize the whole).
Ask, for example: Where are all the places we do forecasts – and are they of high and
trustworthy statistical quality? Where do we need to optimize something? Where else
would forecasts or optimization make a difference? Be sure to cover different types of
decisions and useful anticipations, for example:
9
Successfully Automating Decisions with Business Analytics
You will, of course, want to make your own assessment of the analytical potential of
your business processes.You might also gauge the difference between the current
level of analytical support and the analytical potential of each process.You’ll likely
find opportunities everywhere – the challenge is ensuring that the opportunities, if
exploited, will truly drive business performance and competitive advantage. And keep in
mind that there will still be important roles for ad hoc or “craft” analytics in conjunction
with processes that may not be ready for embedded or “industrial” analytics.
10
Successfully Automating Decisions with Business Analytics
Setting Ambition
It’s useful to categorize your current analytical applications and potential analytical
opportunities – to understand your emphasis and “mix,” to identify gaps and adjacent
opportunities, and to set a realistic level of ambition. Figures 7 and 8 are frameworks
for doing so.
Content Type
Information Insight
This two-by-two matrix is a convenience. Any given analytical application may have
elements of more than one category. However, you can learn a lot about what you
are – and aren’t – doing analytically by using this matrix. And you can explore your
options for turning past information into future insight. The “Ladder of Analytical
Applications,” Figure 8, depicts how analytical capabilities and applications tend to
build upon one another. Going up the ladder entails more sophisticated analytics.
The foundation is good data – is it accurate, consistent, integrated, accessible,
and relevant? Statistical analysis of that data yields useful segmentation (e.g., of
customers, products, and transactions or other business events), which in turn enables
differentiated action (e.g., treating customers differently, or calling upon the right “lane”
in a flexible business process). Incorporating predictive analytics enables a business
to marshal its resources where opportunity is greatest. At the institutional action
rung, differentiation and prediction are embedded in ongoing business processes and
accomplished automatically. The top rung is the domain of real-time optimization,
where the process adjusts on the fly to maintain optimal business yield. As with any
such classification, the higher rungs tend to be less populated.
11
Successfully Automating Decisions with Business Analytics
Real-Time Optimization
Optimal
response
embedded in
real-time Institutional Action
process
Prediction and
differentiated
action embedded
Predictive Action in process
Predictions of
response by
target/segment Differentiated Action
Different
approaches
for different
Key Targets/Segments targets/segment
Key targets and
segments defined
Data in Order
Well-defined
common, clean, and
integrated data
The Ladder is a general pattern, not a set sequence. But within any major process
area, it’s enlightening to ask: How high have we climbed? How high can we
climb? How high does the market want us to climb? Have we compromised our
performance by skipping or only partially covering a rung? Can we gain competitive
advantage by moving up a rung?
12
Successfully Automating Decisions with Business Analytics
• Anticipating whether the business will be able to implement the application and
realize the opportunity.
• Does the process call for or benefit from consistent decisions, including the
need for exacting policy compliance or significant liability from inconsistently
applied policies?
• Is the process high-volume and performed often, and are there significant cost
savings from speeding decisions, eliminating steps, or selecting the best process
variation?
In the rest of this section, we discuss three ways in which the qualification of a
business analytics application differs from that of more conventional and predictable
projects or initiatives.
13
Successfully Automating Decisions with Business Analytics
• Enterprise. Does the process have the cross-functional or cross-boundary scope Solution
to make a difference in overall business performance, including financially?
Predictive modeling and data mining
techniques to better understand each
• Leadership. Does the process have a capable owner who can work across the
customer’s buying profile and identify
affected functions and who has the backing of top management?
the right offer at the right time for each
customer.
• Targets. Will embedded analytics drive the performance of this process to
a level that makes a difference in overall business performance, including
Results
financially? Do we have the metrics to track these results?
Hallmark now uses insights from their
• Analysts. Do the analysts have sufficient analytical ability as well as process or analytics initiatives — such as what
domain knowledge? Who can bridge any gaps? Will process performers be each customer likes to buy and how
prepared to work in an analytics-assisted manner? they prefer to be communicated with
and rewarded — to customize the
Assess the readiness of the process and the organization in all five of these areas. content and timing of direct marketing
You may need to fill gaps in order to have the foundation for succeeding with vehicles. Hallmark marketers have
process analytics. also learned things like the best time
to advertise and for what holidays,
and which factors influence whether
a customer will visit a Hallmark store.
Motivation and Momentum The bottom line for Hallmark is more
value from customer data leading to
Organizations face a variety of challenges in generating management motivation and better decisions and, ultimately, better
organizational momentum for business analytics initiatives. Figure 9 summarizes three ROI on marketing spend.
common scenarios:
• Sometimes there’s a well-defined business problem or improvement
opportunity in a process that lends itself to analytics. Examples include logistics
at Wal-Mart and revenue management at Marriott. Here the target may be clear,
but management motivation may not be high enough if the problem is owned
by functional managers. You may need to elevate sponsorship, especially if the
analytical solution will be cross-functional.
14
Successfully Automating Decisions with Business Analytics
• Sometimes (often as the result of “big picture” thinking) top management senses
or recognizes a strategic business opportunity that depends on analytics. As we
mentioned earlier, this analytical opportunity may be sensed in only an intuitive
way. This may seem the ideal scenario for generating momentum; however, any
delay may frustrate management’s expectations and drain momentum. And
delay is common, especially when the opportunity needs further definition and
qualification, the structure and management of the process are weak, or key
assets (especially data) are not ready.
15
Successfully Automating Decisions with Business Analytics
Experimentation
You may, for example, be confident that better decisions on the timing and priority
of order fulfillment will lower inventory and logistics costs and raise customer
satisfaction. But you don’t know how much effect a better decision system will have
on each of those business goals. The first implementation of the decision system may
just give a foretaste of the business value generated as the model is refined over time.
And you may well discover new goals and new value over time.
Thus, the final major question in qualifying a process analytics opportunity is “Can we
handle a business experiment?” Do we have the right mindset and expectations, both
among senior management and across the organizations affected by the analytical
application? Are our project approval and management methods flexible enough
to guide a business experiment through to conclusion, one way or the other? Are
sponsors afraid to change course or pull the plug because they fear losing face?
If the company’s project approval process (e.g., with strict financial hurdle rates),
project management process (e.g., based on sequential milestones), and management
mindset assume lockstep predictability in business projects, then an analytical initiative
of any scope or scale is destined to run afoul of “the system” and fail – no matter
how good the concept, analysts, and technology behind it. In a situation like this, you
may be able to engage top management to break through the project approval and
management obstacles, and then “cover your flank” as “the system” inevitably tries to
reassert itself and undermine your efforts. Otherwise, you have to start small with
process analytics – a local application entirely within the domain of an enthusiastic
management sponsor.
16
Successfully Automating Decisions with Business Analytics
Summary
A systematic approach to identifying and qualifying opportunities for embedding
analytics into business processes will result in more focused efforts, fewer false starts
and much greater chances of successful implementations. That approach should
encompass the following steps:
• Find analytical opportunity through “big picture” thinking – keeping tabs on what
others in your industry are doing, using business decomposition and analysis
frameworks to assess where performance can or must improve, and exploring
your hunches about the business, what makes it tick, and where the next
breakthrough may await.
17
Successfully Automating Decisions with Business Analytics
About nGenera
About Intel
Intel (NASDAQ: INTC), the world leader in silicon innovation, develops technologies,
products and initiatives to continually advance how people work and live. For almost
15 years Intel and SAS have collaborated to solve real business problems in the most
cost-effective manner. Working together, SAS and Intel have optimized SAS Business
Intelligence and analytics applications to take full advantage of the performance,
scalability and reliability of the latest Intel® processors.
About SAS
SAS is the leader in business analytics software and services, and the largest
independent vendor in the business intelligence market. With innovative business
applications supported by an enterprise intelligence platform, SAS helps customers at
45,000 sites improve performance and deliver value by making better decisions faster.
Since 1976 SAS has been giving customers around the world THE POWER TO KNOW.®
18
www.sas.com www.intel.com www.ngenera.com
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are
trademarks of their respective companies. Copyright © 2008, SAS Institute Inc. All rights reserved. 103717_516697.1108