Big data analytics promises to boost customer centricity and profitability for financial services firms, especially when applied to market research, customer segmentation, product testing, product development and customer service.
The Work Ahead in Intelligent Automation: Coping with Complexity in a Post-Pa...
The Economic Value of Data
1. • Cognizant 20-20 Insights
The Economic Value of Data
Part 1 in a multipart series
Big data analytics promises to boost customer centricity and
profitability for financial services firms, especially when applied to
market research, customer segmentation, product testing, product
development and customer service.
Executive Summary type of data in the right format for analysis; too
much data, as some organizations have learned,
Big data has gained significant influence in recent
may not always be beneficial. We also advise
years and is rapidly transforming the business,
companies to start small and take manageable
operations and technology landscape for a
steps toward incorporating big data analytics into
myriad of industries. Early adopters — particularly
their operating models.
in retail and consumer products — have already
derived significant business insights from big This white paper is the first in a series that
data management best practices, such as analysis presents our perspective on the economic value
of both the growing pools of structured trans- that can be derived from big data analytics by
actional data from operations systems and the financial services companies. Subsequent white
unstructured and semi-structured data generated papers will cover specific functional areas across
by social media interactions. the financial services spectrum in which big data
analytics can have a significant impact.
According to a recent report by International
Data Corp., big data is the next essential business
From the Beginning
capability and a foundation for the intelligent
economy. According to the report, the worldwide Coined by McKinsey & Co., the term “big data”2
big data market is expected to grow from $3.2 describes large datasets that cannot be captured,
billion in 2010 to $16.9 billion in 2015, a compound managed or processed by commonly used software
annual growth rate (CAGR) of 40%.1 tools within a reasonable amount of time and at a
reasonable cost. According to IDC, about 90% of
Investments in big data solutions have helped available data today has been generated in just the
enterprises achieve customer centricity and last two years. In fact, IDC estimates that:
material gains in pricing and profitability. This
whitepaper posits potential opportunities that big
• Data volumes are growing at 50% per year, or
more than doubling every two years.
data analytics can create for financial services
companies, citing specific business opportunities • Machine-generated data is projected to rise
and benefits. Our empirical experience suggests from today’s 200 exabytes to 1,000 exabytes
it is critical to generate the right amount and by 2015.
cognizant 20-20 insights | february 2013
2. • Multi-structured data content is the primary can now be aggregated and analyzed to identify
driver of new data. 80% of this new data is escalation and complaint triggers, understand
digital, which is complex to analyze in its native fraud patterns, manage alerts, reduce credit risk
structure. and build social media dashboards. These devel-
opments can help financial institutions tailor their
• Digital
data is growing at 62% annually vs.
products and build strategy roadmaps aligned
structured data at 22%.3 This explosion of data
with customer expectations. Effective use of
coupled with the growth in social networking
big data will be a key driver for competition in
and virtualization, has introduced unprece-
financial services, and companies that use data
dented opportunities for companies to better
more effectively will secure an edge in the mar-
connect with consumers and
ketplace.
Most companies understandwhere the markets
as well as
their sentiments,
have yet to find are heading. Retaining customers and satisfying consumer
expectations are among the most serious
precise answers to Most companies have yet to challenges facing financial institutions.
the challenges posed find precise answers to the Sentiment analysis and predictive analysis are
by fast-changing challenges posed by fast-
changing consumer demands;
two techniques that they can use to effectively
address these and other key challenges.
consumer demands; many lack the ability to process
many lack the ability data in near real-time or Capturing Customer Feedback Through
Sentiment Analysis
to process data in convert interactions into trans-
actions. Big data analytics, Consumers today are just as willing to share
near real-time or therefore, has become one of their thoughts on social media platforms such
convert interactions the most frequently discussed as Facebook and Twitter as express them to
into transactions. topics for many business
leaders.
a customer service representative over the
phone, Web site or in person. When captured and
managed, such information can provide valuable
Emerging big data tools provide companies with
insights into what customers are thinking. In
the ability to analyze far greater quantities and
addition, online customer reviews on Web sites
types of data in a shorter span of time. It includes
such as Amazon and Yelp are fast emerging as
structured datasets, such as information stored
major influencers of vendor/provider selection
in databases; semi-struc-
and purchasing behavior. This means financial
tured data like XML files and
Snippets of RSS feeds; and unstructured services institutions must carefully review this
proliferating stream of unfiltered content to
unstructured data datasets, such as images, gauge customer expectations and opinions on
can be interpreted and videos, text messages, e-mails product offerings and then act accordingly.
analyzed, delivering and documents. New technolo-
gies can help uncover insights Traditionally, companies have collected consumer
insights that can hidden within these large feedback using survey and focus group results.
determine likely datasets. While retailers and These tools may gauge consumer sentiment, but
consumer response technology companies such as
Google, Walmart, Amazon and
they may not necessarily capture emerging trends
or hidden insights, particularly on a real-time
(both favorable Sears have made significant basis. A negative opinion of a bank’s offering can
and unfavorable) to developments on this front, potentially lead to dramatic customer churn. For
decisions made the doorsthe financial services
open for
are just starting to example, in September 2011, Bank of America
announced its decision to charge customers a
by the bank. industry, which stands to gain monthly debit card fee. Three days later, the bank
significant advances in areas withdrew the decision after a customer uproar
such as market research, customer segmenta- and threat of attrition. The reversal occurred after
tion, product testing, product development and customers petitioned the bank and mobilized to
customer service. close their accounts and take their banking and
investment business elsewhere.4
For example, text captured from credit applica-
tions, account opening interviews, call center Sentiment analysis tools aim to capture customer
notes, mortgage application notes, social media feedback from social media platforms and
chatter and other customer service interactions customer service interactions, among other
cognizant 20-20 insights 2
3. sources, and help banks evaluate the potential service features. This can help banks generate
impact of such decisions. Sentiment analysis customer “wish-lists” and incorporate these into
enables organizations to associate words used their product roadmaps.
in unstructured communications and tie them
to consumer emotions and sentiment on a topic. Sentiment analysis can also help banks reward
These findings can serve as key inputs into customers effectively. This is extremely important
strategic decision-making. across the industry because account switching
costs are relatively low and customer churn is
The idea is to use technology to create codes a major challenge. By examining customer con-
that analyze the Web and provide insights into fidence indices that are
consumer sentiment on a much larger scale and driven by specific data
at a much faster rate than the findings revealed elements (product, func- By examining customer
by surveys or focus groups. Snippets of unstruc- tionality, content and confidence indices that
tured data can be interpreted and analyzed, price), banks can judge the
delivering insights that can determine likely mood of the market and
are driven by specific
consumer response (both favorable and unfavor- decide how to best reward data elements (product,
able) to decisions made by the bank. their customers. Success- functionality, content
ful execution drives loyalty
Consider the following customer scenarios and and also attracts new cus-
and price), banks can
statements: tomers. Figure 1 illustrates judge the mood of the
how banks can effective- market and decide how
• “ABC Bank’s small business offering is useful ly satisfy a disgruntled
for new businesses and entrepreneurs. The
customer using the afore-
to best reward their
lack of a same-day payment facility is a downer,
mentioned technique. customers. Successful
though.”
execution drives loyalty
• “The feature to view both business and Although the technologies
personal accounts is really cool, although they behind sentiment analysis and also attracts new
really need to improve their customer service.” are still maturing, many of customers.
the tools and techniques
The sentiment analysis tool would pick up words are advanced enough for financial services insti-
like “useful,” “lack” and “improve” and attach tutions to derive incremental value by under-
contextual meaning to generate graphs and standing customer likes, dislikes and preferences
reports, which can then be used by the bank for product and service improvements. Clearly,
to satisfy customer expectations. Additionally, early adoptors will gain a competitive advantage
reports can be generated to illustrate trends and going forward.
opinions on individual product and customer
Converting Detractors into Advocates
Michael has recently • The nature of the e-mails • The bank sends Michael a
registered several suggests a disgruntled personal note addressing his
complaints with customer customer that is likely to churn. concerns.
care at his bank. The bank recognizes this and
takes immediate action.
• The bank offers to refinance
his auto loan at a much better
• The bank knows Michael has rate, saving him money and
a new car loan. gaining his loyalty in return.
Figure 1
cognizant 20-20 insights 3
4. Using Predictive Analytics to churn or favorable response to a particular
Capitalize on Customer Insights marketing campaign. For example, our work with
Merchant Rewards International, a provider of
Customers across the globe are increasingly
credit card processing services, indicates a higher
demanding simple, fast and inexpensive means
response rate for offers aligned with previous
to conduct both financial and purchasing trans-
transaction behavior and buying propensity.
actions. However, consumer needs are becoming
more diverse and unpredictable, Predictive analytics can help banks build models
Institutions can placing tremendous pressure based on customer spending behavior and
create pre-defined on companies to fulfill them.
Ongoing economic challenges,
product usage to pinpoint products and services
that customers might find more useful and that
profiles, thereby accelerating globalization and financial institutions can deliver more effectively.
revealing a history of provider choice means financial Such a model can help banks develop an efficient
higher fraud volume services firms must meet and
exceed traditional expecta-
cross-sell offer, helping them increase their share
of wallet, garner loyalty and increase profitability.
through purchase tions. While failing to respond
types and to dynamically changing expec- For example, profiling technology can help credit
ticket sizes. tations is problematic, a larger
challenge is correctly predict-
card companies identify transactions, cardhold-
ers and merchants that exhibit a high probabil-
ing consumer needs and desires ity of fraud. Institutions can create pre-defined
and responding, just in time, with the right set of profiles, thereby revealing a history of higher
products and services. fraud volume through purchase types and ticket
sizes.
Predictive analytic techniques can be used to mine
large amounts of historical data and determine Furthermore, predictive analysis can identify
the likely occurrence of events in the future. aberrant behavior patterns and help financial
By querying, visualizing and reporting these institutions prevent fraud. Collecting data from
datasets, companies can generate actionable multiple sources, such as Web behavior and point-
insights. Changing data over time can illuminate of-sale inputs, and correlating it with aggregated
behavioral and transactional patterns that can data compiled from other financial services
help with move-forward decisions on product and firms by third-party providers, can help banks
service strategies. and brokers detect fraud earlier than existing
approaches. Big data analytics not only helps
Regression and response models are among the financial institutions preserve the long-sought-
techniques that financial institutions can use to after “instant transaction user experience,” but it
determine, for example, the likelihood of customer can also safeguard them against fraud.
Unleashing Machine Intelligence
Michael uses his credit • The predictive analytics system • The system allows the
card to perform an online determines Michael’s location, transaction to go through
transaction. time, transaction category and if the generated score
merchant. probabilistically
determines that the
• The system then compares purchaser is actually
these details with Michael’s
Michael.
past purchase behavior and
calculates a score.
Figure 2
cognizant 20-20 insights 4
5. A good example is the ongoing refinement of talent to accelerate their analytics efforts, as they
neural network technology5 to assess whether often lack internal expertise or cannot afford to
a credit card transaction is being performed stretch existing resources. In some cases, instead
by the real cardholder or someone committing of hiring and motivating talent internally, they are
fraud. The transaction is scored against a pre- engaging third-party providers to supply talent
defined profile, and if the score passes an estab- on an “as needed” basis.
lished cutoff, it is approved; otherwise, it is held
for a fraud check. Banks have used this type of Looking Ahead
artificial intelligence technology since the early Big data analytics can help financial institutions
1990s to perform pattern recognition and spot derive significant benefits by increasing customer
fraudulent transactions. However, big data tech- satisfaction, retention and expansion through
nologies make the process faster and more cost- more effective cross-selling and improvement of
efficient, accurate and robust. their fraud and risk management capabilities. The
economic value of data will be
Financial institutions can also create ‘”predictive realized only when financial
scorecards,” which can help determine the institutions fully endorse As risk managers
likelihood of customers defaulting on payments big data analytics and invest frequently say,
in the near future. Among the parameters to in innovation. Although the it is better to be
consider are late utility bill payments, late car possibilities are endless,
insurance payments, increases in purchases numerous challenges must approximately right
compared with monthly averages and listening be addressed before the than precisely wrong.
and learning from relevant social media conver- benefits can be fully realized.
sations.
In a special report in The Economist, author
As with sentiment analysis, additional research Kenneth Cukier reveals that the recent global
and development is required to improve the financial crisis sheds light on how banks and
accuracy and effectiveness of predictive analytic ratings agencies relied on models requiring
techniques. However, when deployed strategical- vast volumes of information but failed to reflect
ly, these tools can help banks gain a significant financial risk in the real world.6 Therefore, to
advantage in a competitive macro-economic envi- capitalize on big data analytics’ opportunities and
ronment. realize significant business value, it is advisable
for financial institutions to start small and grow
Areas such as delinquency propensity, loss
gradually. Firms must find the right balance of
mitigation and cross-sell/next-best-offer scripting
required information and desired insight. As risk
are all specific areas offering a solid business case
managers frequently say, it is better to be approx-
for use of analytics techniques. Financial institu-
imately right than precisely wrong.
tions are making investments and hiring outside
Footnotes
1
“Worldwide Big Data Technology and Services 2012-2015 Forecast,” IDC, March 7, 2012,
http://www.idc.com/getdoc.jsp?containerId=prUS23355112#.UQwxhuTAeE4.
2
“Big Data: The Next Frontier for Innovation, Competition and Productivity,” McKinsey Global Institute,
May 2011, http://www.mckinsey.com/insights/mgi/research/technology_and_innovation/big_data_the_
next_frontier_for_innovation.
3
”Worldwide Big Data Technology and Services Forecast,” IDC.
4
Tara Siegel Bernard, “In Retreat, Bank of America Cancels Debit Card Fee,” The New York Times,
Nov. 1, 2011, http://www.nytimes.com/2011/11/02/business/bank-of-america-drops-plan-for-debit-card-fee.
html?_r=0.
5
Donald F. Specht, “Probabilistic Neural Networks,” ScienceDirect, 1990,
http://www.sciencedirect.com/science/article/pii/089360809090049Q.
6
“Data, Data Everywhere,” The Economist, Feb. 27, 2010, http://www.emc.com/collateral/analyst-reports/
ar-the-economist-data-data-everywhere.pdf.
cognizant 20-20 insights 5