Sustaining analytics

Deliver hindsight, insight,

and foresight
Analytics: Fad or foundation?
Knowledge is power, says the old axiom. Clich as it
may be, it has never been more true, especially when it
comes to the grueling effort most companies face today
in gaining and maintaining a competitive advantage.
Enhanced access to information can lead to better decision
making, improved performance, and, ultimately, a better-
looking bottom line. Such erudition doesnt come easy,
however; it takes a keen understanding of the business
environment coupled with an information technology
infrastructure that facilitates the timely, careful acquisition
and analysis of information.
The emergence of advanced analytics over the past three to
ve years has been a great boon for many organizations in
their quest to enhance their ability to capture information
that is meaningful and actionable. However, as with
any quickly emerging phenomenon, there are questions
surrounding the staying power of analytics. The biggest
question is: Is analytics a fad or is the mastery of analytics
techniques and technology foundational to success?
The answer to that question will depend on how
companies implement and maintain analytics solutions.
Analytics doesnt happen in a vacuum. Companies looking
to enhance their analytics capabilities must take a holistic
view of the data resident in their information systems.
Effective information management is a foundational key
to successful analytics. Deriving hindsight from business
intelligence and performance management systems is
important to understand what has happened, but insight
into what may happen is even more crucial.
High strategic value and thus higher return on
investment (ROI) on information technology (IT)
investments comes from true analytic insight. Knowing
what tools and technologies to apply to yield real business
results is critical to an organizations analytics journey.
Deeper analytics pushes into the organization, embedding
the capabilities, the outputs and inputs, at the strategic and
operational levels. Knowing that, the key question then
becomes, How do we sustain what weve built here?
Sustaining analytics and making it foundational to the
organizations continued success is a journey. It is an
ongoing effort to excel, to make the most of the IT
resources, and to harness the enormous power of analytics
to better understand and grow the business. It requires a
disciplined approach that will take no less than the best and
brightest resources the organization can muster to transform
itself and gain the power that comes with true knowledge.
Weaving analytics into the ber of the organization
Theres no doubt analytics is on the forefront of the
C-Suites agenda these days. According to a recent Gartner
2
Group study, Leading in Times of Transition: The 2010 CIO
Agenda, improving analytics capabilities is among their top
three priorities, and its their number one technical priority.
1
The process of delivering analytics business results is one of
continuous improvement. Starting anywhere in the analytics
cycle, an organization can quickly begin to address its specic
needs. Deep analytics pushes insights to frontline decision
makers by embedding capabilities and outputs deep in the
processes of everyday work. However, that process has to be
wrapped with serious change management for it to stick.
Those organizations governed by a gut-driven
information culture must instead embrace a fact-driven
mentality. Intuition should be replaced by a fact-driven
approach to decision making in which actions are backed
by data, not instinct. It is also imperative to set and
communicate an enterprise analytics strategy and integrate
into operations. Predictive models must be integrated into
processes, technology, and the workforce. Performance
should be measured and managed with metrics and
processes, via analytical applications. Finally, analytics must
also be linked to high-impact processes. Business processes
especially those deemed most critical to business
performance must be driven by analytics.
The following gure shows how analytics is interwoven
into high-impact business processes:
Figure 1: High-impact processes that link to analytics
1 Leading in Times of Transition: The 2010 CIO Agenda. Gartner
Group, January, 2010, p. 5.
Some of the key high-impact processes are:
lroduct menegement. product pr|c|ng, product
protability, portfolio optimization
Customer menegement. ||fet|me ve|ue, customer
segmentation, loyalty, customer experience,
protability, churn
Serv|ce/operet|ons menegement. customer
exper|ence, cepec|ty p|enn|ng/demend forecest|ng,
cep|te| expend|ture, performence, |ee|ege/shortfe||,
workforce effectiveness
Lnterpr|se menegement. freud, revenue essurence, esset
utilization, security, collections, advanced forecasting
Supp||er/pertner menegement. vendor efc|ency,
contract compliance, vendor optimization
Mer|et/se|es menegement. chenne| opt|m|zet|on,
campaign performance, up selling, cross-selling
Linking analytics to high-impact processes can push analytics
capabilities even deeper into the organization, from
business executives to the front lines. No matter where an
organization enters the cycle of business analytics analytics
insight, performance optimization, information management
the challenges of execution and adoption are similar.
The principles of sustainable analytics
Organizations that have mastered applying analytics to
support enterprise decision making tend to focus on seven
analytics principles as they strive to accelerate workforce
adoption and capture more value.
Start where you are
The rst principle of sustainable analytics is understanding
the current state, and picking a starting point for
implementation. An honest self-assessment is the rst step
in understanding current capabilities and the gaps that
will need to be closed to get more value from analytics
investments. Think in terms of both technical capabilities
and organizational depth. For example, is data aggregated
and standardized? Is there an enterprise data warehouse?
How robust are organizational business intelligence
capabilities? Is intelligence really embedded into many
parts of the business? On the organizational side, is there
a data governance structure in place? Has the organization
bought into data governance efforts? Is there a culture of
accountability for data quality?
Advanced
analytics
Customer
management
Product
management
Service/
operations
management
Enterprise
management
Supplier/
partner
management
Market/
sales
management
3
Its not necessary to wait until there is an improved IT
infrastructure in place to begin to analyze information.
Because of the semantic and in-memory analytics capabilities
inherent in of most of the analytics tools today, capabilities
can be added as needed. The key then becomes to assess
where an organization is in the sustainable analytics journey
and begin to develop those areas that are the most ready.
Once this assessment is complete, the next step is to begin
to ask the business questions that really matter.
Ask crunchy questions
Shareholder value can be improved when activities that drive
value are recognized and effectively managed with indicators
that provide lagging, leading, and predictive information
that is based on reliable data. Getting that information isnt
always easy, though. Its essential to know which questions
to ask about the business. In fact, knowing which questions
to ask where the most relevant and valuable insights
are likely to be found is the foundation of any analytics
initiative. Asking relevant questions requires collective
intelligence that spans industry, function, and technology.
Knowing which questions to ask will deliver hindsight
into what happened, insight into what is happening, and
foresight into what may happen. That knowledge can
facilitate improved awareness of real business issues, and it
can result in the realization of higher value from IT systems,
thus producing a higher ROI on IT investments.
Circular as it may sound, asking the right questions
starts with asking the right questions. Using their overall
corporate strategy and business processes as a guide,
organizations rst must ask:
vhet dr|ves ve|ue`
vhet quest|ons need to be es|ed`
ow do we meesure up ege|nst our goe|s, end
against the competition?
vhet |nvestments w||| dr|ve the requ|red return`
vhet |nformet|on do we need to enswer these
important questions?
Once these questions are answered, then the real work
starts. That work begins with asking questions with
answers that matter to the success of the organization
the crunchy questions. Linking the answers to those
questions to the overall organizational strategy, then
linking the strategy to specic industry pressures, is key.
Most orgen|zet|ons espec|e||y over the pest decede or so
have been forced to operate in a volatile environment.
In such chaos, those crunchy questions may change from
time to time. For example, several years ago, when the
capital markets were crashing and the economic situation
for most companies was precarious at best, one crunchy
question that many CEOs had was, How do I re-forecast
my nancials to reect such an unpredictable situation?
Another question was, How do I adjust my workforce to
meet current needs? Still yet another was, How is my
supply chain, my inventory management affected? Being
ready for those crunchy questions and ready for those
questions to change on a dime is essential.
Nonexistent Developing Dened Advanced Leading
Advanced
analytics
|nds|ght dr|ven
No mer|et |ns|ght
No ene|yt|c
applications
lunct|one| |eve|
/cross few dome|ns
Le||ver to|en resu|ts
/ne|yt|cs |so|eted
Compet|t|ve
advantage analytics
Lxecut|ve support
/ne|ysts d|sperete
/ne|yt|cs |so|eted
/ne|yt|cs |ey
/ne|yt|cs |ncent|ves
Operet|one||zed
Levereg|ng
analytical tools
/ne|yt|csdr|ven
strategy
C|obe| ene|yt|cs
/ne|yt|cs |ntegreted
/ne|yt|cs me|nstreem
Performance
management
Lxce| besed
budget ene|ys|s
lrocess|eve|
decisions
Mer|et dr|ven
l|nence hes lM s||||s
/sp|ret|on for more
Scorecerds too|s
Klls dened
L|n|ed Klls
Soph|st|ceted s||||s
Stenderd|z|ng lM
lntegreted lM
Ve|ue||ned Klls
/||gned |ncent|ves
lntegreted lM
lM |s ene|yt|cs dr|ven
lM COL d|ssem|nete
skills
lred|ct|ve
mode|||ng/lM
Business
intelligence
bl ||m|ted to report|ng
bl s||||s bes|cs
bl s||||s |n l
Lete unre||eb|e
bl |so|eted
Lv esteb||shed
bl stretegy dened
bldr|ven stretegy
Led|ceted bl teem
Owners of bl
bl governence
bl/b/ s||||s |n l
bl erch|tecture
lu|| |eve| of bl too|s
Stretegy dr|ven
bl/b/ resources
Stenderd|zed bl too|s
lnformet|on dr|ven
Data
management
No governence
L|tt|e dete que||ty
No MLM
No l bl/LM s||||s
L|m|ted LM s||||s
Leve|op|ng LM
processes
Recogn|ze need
LM governence
LlM erch|tecture
Lete que||ty/MLM
Led|ceted LM teem
lu|| dete governence
LM teem |ntegreted
MLM |mp|emented
Meture LM too|s
|ghque||ty dete/MLM
S||||ed LM resources
lbl/LM governence
Stenderd LM
Figure 2: Assess current capabilities
4
Listen carefully
Understanding what signals matter most is the next
fundamental principle of sustaining analytics excellence.
Business is not conducted in a vacuum. Almost
innumerable internal and external forces act to inuence
events and decisions based on those events. Each event
that happens leaves an imprint of some kind, somewhere,
in the business environment.
Signals are everywhere
As a hypothetical example, consider a large global
conglomerate that dabbles in everything from energy
to manufacturing. This company has just experienced a
catastrophic incident at one of its remote worksites. A
mammoth explosion has decimated multiple structures
at the site, killed several employees, and seriously injured
scores of others.
Menegement |s |n cr|s|s mode, scremb||ng to nd out
what happened and more importantly why. The
company is excoriated by the media and the international
community. The companys reputation takes an enormous
hit, and it is exposed to enormous liability. Could this have
been prevented, or was it just a random occurrence that
was completely unpredictable?
In most accidents, there is no single cause. Instead, there
are many small events that occur in precisely the right
sequence or combination to create a cataclysmic failure.
This is certainly the case here. These events are signals and
they can come from almost any source.
Months eer||er, sefety |ssues et the s|te were revee|ed but
not edequete|y eddressed. vor|er more|e et the s|te wes
low due to isolation and a harsh management style. Some
workers had posted comments about conditions on various
soc|e| med|e s|tes. vor| sefety procedures were not rout|ne|y
followed, and equipment was not regularly or closely
inspected for damage, although it was under tremendous
stress. Inspection reports that were completed were lost.
vorse, superv|sors hed de||berete|y d|seb|ed over|y
sensitive alarm systems so the alarms would not trip at all
hours of the night. Finally, executive management rarely
performed in-depth analysis on reports from the site
trusting site management to deliver correct information.
Early one morning, a critical equipment failure occurred,
setting off a chain reaction of events that resulted in a
massive, deadly explosion.
All of these seemingly unrelated events combined to create
the catastrophe. Could they have been detected sooner
and connected? Could the accident have been avoided?
Perhaps. Each event was a signal that was recorded or
should have been by one or more of the companys
information systems.
Asset tracking systems should have caught the equipment
issues. Risk management should have been aware of the
safety issues. Human resources and executive management
should have been apprised of and acted on morale
issues. These are but a few of the information systems
that would have been involved. Unfortunately, there was
no capability in place to allow management to combine
seemingly unrelated events into a cohesive picture that may
have signaled a disaster in the making or at least made
clear that multiple, critical issues needed to be quickly and
effectively addressed to prevent a massive failure.
Deciphering the signals
Admittedly, most organizations rarely face this kind of
disaster. However, they do face a never-ending stream of
events that affect business performance, both positively
and negatively. The ability to decipher and integrate
these signals into a coherent reection of performance is
vital. Signals come from myriad sources, both externally
and internally. External signals can come from market
conditions, news media, social media, wisdom of the
crowds, etc. Internal signals are captured by transactional,
business intelligence, asset tracking, risk management, and
nancial management systems, as well as the ever-growing
volume of unstructured data from e-mails, customer
interactions, surveys, video the list goes on and on.
The gure below depicts how events from multiple internal
and external sources are interrelated and produce signals
that must be detected and analyzed in light of other events
that are occurring, or are expected to occur.
5
Figure 3: Signals come from many sources external
and internal
Critical information from these signals must be organized,
aggregated, standardized, and accessible to answer critical
business questions, provide foresight, insight, and measure
success. Once the information has been processed, the
next step is to apply business process analytics, asset
intelligence tools, and exploratory data analysis to monitor
decisions, strategies, as well as transition optimization
and risk management strategies. Using these tools and
techniques can facilitate the aggregation of seemingly
independent events into a coherent picture of whats really
happening across the business landscape.
Accelerate insights
To accelerate the process of deciphering signals and
answering critical business questions, it is essential to
automate information delivery to processes and people,
and to automate responses as much as possible, so that
decision making is sped up and action can be taken with
certainty, at the lowest cost.
Pricing Automated
One key business function that is rapidly being automated
by leading companies is pricing. There are a number of
ways to introduce automation into the pricing function:
lr|c|ng stretegy. Compen|es cen deve|op pr|ce
optimization models to allow for more careful and timely
pricing and incorporate data from pricing agreements
to allow for more granular price setting. They can then
implement an enterprise-wide pricing software tool to
support pricing processes and align and integrate the
tool with the organizations ERP system
/||gn|ng pr|c|ng opt|m|zet|on w|th compenset|on.
Companies can also develop compensation plans for
the sales organization that are aligned with protability
metrics. They can then implement those plans by revising
the sales force commission structure to be based on a
predetermined, but adjustable mix of volume and margin
l|e|d execut|on of pr|c|ng stretegy. o tre|n the
sales team, companies can utilize a high touch
implementation method to explain pricing and
protability metrics and how the team will be incented
to meet them. They can then develop a list of
performance review activities to allow for measurable
progress in meeting goals
Customer communication Enhanced
Automation can also play a role in how companies choose
the most effective channels for customer communication.
As an example, consider a large retailer that has an
established online and social media presence and executes
most of its marketing communication electronically. To
facilitate more effective customer communication, the
retailer has built an automated system that monitors each
type of customer contact and message that it sends. Lately,
the retailer has undertaken a project to understand which
types of communications are most and least preferred
by its customers. Analysts have developed an automated
tracking system to analyze user actions upon receiving
e-mails, or upon encountering an online advertisement.
Signal
analysis
????
Action
Predictive
markets
Internet
analytics
E
-
d
i
s
c
o
v
e
r
y
/
u
n
s
t
r
u
c
t
u
r
e
d
A
s
s
e
t

i
n
t
e
l
l
i
g
e
n
c
e
I
n
t
e
r
n
a
l

s
t
r
u
c
t
u
r
e
d
E
x
t
e
r
n
a
l
s
t
r
u
c
t
u
r
e
d

a
n
d

u
n
s
t
r
u
c
t
u
r
e
d
Equipment
facilities
energy
Excel
operational
systems
E-mail
videos
Economic
geological
climate
Wisdom of crowds
Social Web analytics
6
Specically, the system tracks whether recipients open
e-mail messages or mark them as spam, or whether or not
they click through an ad in order to view the advertised
site. The system will help retailers marketing function infer
the likelihood of whether a particular communication is
high value, or whether it is ignored by users.
For example, if too many recipients mark a message as spam,
the system would simply stop sending that type of message.
In the case of advertisements, the ad developer is notied,
also through an automated system, of the low click-through
rate, and they can be encouraged to develop an ad that
users might respond to more favorably. Utilizing this type of
automation in the marketing function helps management
refocus resources to more critical tasks, because they no
longer need to micromanage every customer interaction.
Enhance visualization capabilities to engage users
Automation is key, but if users cant interact well with the
systems that deliver information, they wont use them,
and the analytics effort will ultimately fail. Accordingly, to
promote sustainable analytics, an engaging user
env|ronment |s essent|e|. ve||des|gned dete v|sue||zet|on
capabilities and interfaces can let business users easily access
actionable information, or drill down for more information.
To foster a user-friendly analytics environment, its crucial
to involve the user community in the design process. They
have the eld experience to know what information they
need, and further, they most likely have good ideas on
how to present that information to facilitate ease of use.
As an example, consider the risk management function
at a major health insurer. Analysts are tasked with
gathering, and presenting information on, provider costs
so that management can ferret out potential suspicious,
or fraudulent providers. The choice is to present rows of
numerical data, or to present that data in a graphical format.
Its not difcult to discern which format is more engaging.
The chart below is a scatter plot depicting the number of
patients across the cost of a claim. There are easily visible
patterns in the chart, and there are outliers that call for
more analysis. In the analysis of the outlier data, there are
also patterns and some clear anomalies as well.
Figure 4: Enhanced user visualization yields clear results
Clustering and anomaly detection
identies health care providers that
consistently claim higher charges
$0 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000
Cost/claim
3,500
3,000
2,500
2,000
1,500
1,000
500
0
N
u
m
b
e
r

o
f

p
a
t
i
e
n
t
s
Specialty providers
Suspect provider
7
The rst question to ask of this data would be, How do we
explain the outliers? A deeper drill-down into those results
might reveal that those providers were performing special
procedures that were very expensive, but necessary and,
therefore, well within treatment guidelines and acceptable
performance parameters. Examples of this situation could
include expensive chemotherapy regimens, heart bypass
procedures, or other similar treatments. However, there is
still an outlier to the right of those outliers that have been
accounted for. That outlier is a good candidate for analysis
of suspected freudu|ent ect|v|ty. v|th the dete presented
in a graphical format, its easy to spot both patterns and
anomalies that call for further analysis.
The upshot is that organizations that want to sustain
analytics efforts should understand that the method of
presenting information for analysis is just as important as
the information itself. It must be presented in a format that
engages the user and promotes digging deeper to nd
answers that might otherwise be overlooked.
Build a fact-driven culture
Even the prettiest chart is only as accurate as the facts
behind it, however. Good questions that yield wrong
answers due to inaccurate information can ruin even the
most carefully implemented analytics initiative. To sustain
an analytics program that provides systematic answers and
facilitates enhanced decision making and performance
management, its critical to build an organizational
information culture that is fact driven.
Years ago, it was in vogue to shoot from the hip in
dec|s|on me||ng. Mever|c| CLOs were e|| the rege end
their successes were touted by pundits as a nod to
trusting your gut and making the hard decisions based
on experience and knowledge, rather than often unreliable
facts and gures. For some organizations, this approach
worked for a while. Times have certainly changed. Over
the last decade or so, a wave of accounting scandals,
external threats, increased regulatory scrutiny, and
economic adversity has put a new premium on having
and making decisions according to hard facts.
To operate in todays complex and cut-throat business
environment, many leading organizations are building a
fact-driven analytics culture to drive streamlined execution
of business processes and decision making throughout the
enterprise. There are several steps an organization can take
to build a fact-driven culture:
Cet |eeders who we|| the we||. / ClO or CLO who seys,
These are the inputs I use to make decisions. will go a
long way to facilitate adoption of those inputs by others.
lrect|ce proect|ve obso|escence. One effect|ve wey
to make sure outdated ideas and analytics tools are
eliminated is to formally forbid their continued use.
/s| tough quest|ons. vhen rev|ew|ng dec|s|ons mede by
others, ask to see which analytics methods were used in
the decision making process. Strongly encourage fact-
based inputs as well as assumptions.
Cet out eheed of chenge. No metter wh|ch epproech
to change management is deployed, training related to
expectations, resources, and benets will be critical.
Plan for them.
/dopt one vers|on of the truth. One edopt|on berr|er
to analytics is a fear that the facts delivered arent
reliable. To counter this perception, its often necessary
to develop a system of certication to dene selected
reports and ndings as ofcial and stick with it.
Incent adoption. Once there is a stable analytics
infrastructure in place, begin to reward adoption and fact-
based decision making to incentives and compensation.
Meesure whet metters. Lxpend the enterpr|se dete mode|
for reporting KPIs to measure how many decisions
people are making, as well as their effectiveness.
8
Practice right-t analytics
In building a sustainable analytics environment, its critical
to drive decision making with fact-based inputs, but
its also imperative to get at those facts with analytics
technologies that are the right t for the organization as
a whole. The proliferation of analytics tools on the market
has already outrun the pool of available talent to use those
tools powerfully in most organizations. This leads to two
big challenges: setting the overall analytics strategy and
choosing tools and technologies based on that strategy,
and on challenges and constraints faced.
The gure below depicts how understanding business
challenges and constraints to the corporate strategy helps
in determining which analytics solutions are the right t.
Figure 5: Analytics solutions based on challenges and constraints
To get the right t, its essential to take an in-depth look at
the organization as a whole. Examine budget constraints,
stafng levels, and resource availability for analytics efforts.
Take a deeper look at how well management understands
the business environment. Consider risk tolerance for
making decisions. Develop an understanding of data
privacy and regulatory issues regarding data security.
For example, if an organization is using a data-mining
tool, data security is a big concern. These tools typically
touch the database directly. Access to them, and to the
information they gather, must be strictly controlled. There
are also some countries that forbid the transfer of data
outside the country to analyze it. So, if management needs
Business challenges Constraints Solutions
Efciency
Budget
Stafng
Infrastructure
Licensing
Risk tolerance
Urgency
Security
End users
Cost
Risk
CRISP-DM
Linux
SAS
S/CMM
SQL Server
UNIX
CART
Windows
Oracle
Regression
Solaris
SPSS
Experian
SVM
Rapid
Miner
Clustering
a global view of data, analysts must gure out a method to
perform in-country analysis or to legally link multinational
systems to that data. The bottom line is that privacy
concerns, security issues, global regulations, etc., all need
to be well thought out as they relate to choosing analytics
tools that are the right t.
Finally, its essential to examine the various tool and
technology options that meet the needs and constraints that
have been determined. Looking to the future is important
here. Analytics investments must focus on implementing
tools that t now, and that will be exible and powerful
enough to grow with the organization. Straightforward
business intelligence tools are not necessarily good at
analytics, so it might be necessary to examine more
sophisticated tools and technologies to get the right t.
Looking ahead
Sustainable analytics is a journey, not a project. If
organizations are to truly sustain their analytics efforts,
they must be ready for, and embrace, change. Analytics
must be woven into the very ber of the organization.
Analytics tools and technologies must support, and be
embedded in, high-impact areas that drive value. Its also
essential to listen to all available signals that ow into the
organization on a daily basis and use information from
those signals to deliver hindsight, insight, and foresight to
understand whats really happening enterprise wide.
Menegement must e|so be eb|e to th|n| outs|de the box
when problem solving. They must ask the right questions,
and they must be willing to use new inputs and weigh
facts against gut feelings when making decisions. Finally,
analytics tools and technologies must align with and
support the overall organizational strategic direction and
they must help the organization recognize, confront, and
manage the challenges and constraints that inhibit
growth and performance.
For more information, please contact:
Jane Grifn, Principal
Deloitte Consulting LLP
Atlanta, Georgia
Tel: +1 404 631 2506
E-mail: janegrifn@deloitte.com
Join the analytics discussion at
www.realanalyticsinsights.com
Copyright 2011 Deloitte Development LLC. All rights reserved.
Member of Le|o|tte ouche ohmetsu L|m|ted