Bank(ing) on Data Science

Bank(ing) on Data Science
By embracing data science tools and technologies, banks can more
effectively inform strategic decision-making, reducing uncertainty
and eliminating analysis-paralysis.
Executive Summary
Amid the ever-present big data buzz, some large
global banks have mastered the art of using
data science to improve customer engagement,
revamp products and optimize marketing
outreach, risk management, pricing and ongoing
cost reductions. Meanwhile, others are still trying
to make sense of where these emerging technolo-
gies and techniques fit in. At some point, banks of
all sizes, shapes and forms need to incorporate
data science into their operating models.
The future of banking will be determined by how
well banks use technology to maximize their
accumulated wealth of transactional and interac-
tional data to better understand hidden patterns
of customer behavior. Using these insights, they
can make necessary service improvements and
customize existing offerings to properly align the
right products with the right customers.
To successfully implement data science, banks
need to start small and adopt a structured
approach, based on a strategic roadmap. Banks
that can analyze the data they collect and utilize
it for strategic decision-making will maximize
their competitive advantage; those that cannot
will place their profitability, if not their survival,
at risk.
By understanding data and applying insights
gleaned from customers, partners and employees,
banks can more effectively compete on code and
gain incredible competitive advantage. Companies
such as Google, Pandora, Netflix, Amazon — and
many others — are winning decisively in their
markets because of their refined ability to mine
insight from the digital information surrounding
people, organizations and devices, or what we call
a Code Halo™. When properly harnessed, Code
Halos contain a treasure trove of business value.1
This white paper details the growing importance
of Code Halos in data science and analytics ini-
tiatives. Importantly, it highlights potential areas
of fit, ways to overcome challenges and a recom-
mended implementation strategy for key data
science initiatives.
Banking’s Evolving World
In the aftermath of last decade’s global financial
meltdown, the banking industry is undergoing a
radical transformation due to rapidly changing
consumer behaviors and expectations; more
stringent regulatory guidelines; and a highly
competitive environment with a proliferation of
new channels (mobile banking and social media)
and competitors (nonbanks, such as Paypal and
Google Wallet).
This ongoing transformation, while difficult, is
also opening doors to new opportunities. Banks
must ensure that they can cost-effectively acquire
new customers while retaining existing ones. And
• Cognizant 20-20 Insights
cognizant 20-20 insights | june 2014

2
to expand their reach and profitability, they must
also tighten their focus on the expanding digital
world. Analytics, big data and data science can
unlock a world of new possibilities. With proper
use of data science, banks can better understand
prospect/customer relationships by exploring
ever-changing transactional and interactional
behaviors. New digital marketing technologies,
such as Web sites, e-mail, mobile apps and social
networks, are helping banks better target their
customers and improve engagement. Moreover,
advanced segmentation strategies are helping
them boost their marketing effectiveness by iden-
tifying niches based on consumer behavior.
Growing Importance of Data Science
The goal of data science is to extract hidden
insights and knowledge from data. In our view,
the key word is “science,” since, done properly,
data science requires a systematic study of obser-
vation, backed by proven scientific techniques.
Data science builds on elements, techniques
and theories from many fields, including signal
processing, mathematics, probability models,
machine learning, computer programming,
statistics, data engineering, pattern recognition
and learning, visualization, uncertainty modeling,
data warehousing and high-performance
computing. The exponential growth of data, par-
ticularly unstructured data, makes big data an
important aspect of data science. Every day, 2.5
exabytes of data are created; just one exabyte
is equal to 50,000 years’ worth of DVD-quality
video.2
For years, financial institutions have leveraged
customer insights gleaned from systems of record
to manage risk and fraud, as well as to improve
product development, marketing and customer
communications. Today, new and enhanced
technologies, coupled with the availability of a
vast pool of structured and unstructured data,
allows for real-time, multichannel decision-mak-
ing processes that can save money and increase
revenues.
Many banks are just beginning to consolidate and
utilize the internal data elements at their disposal,
such as debit and credit transactions, purchase
histories, channel usage, communication prefer-
ences, loyalty behavior, etc. And when it comes
to big data, banks have collected large amounts
of information from a variety of sources, such as
transaction details and spending behaviors. The
addition of newer sources, including Web server
logs, Internet clickstreams, social media activity
and mobile-phone call details, has opened the
floodgates on the data sets that can be mined for
meaning.
However, this is easier said than done, as these
data sets come in a variety of structured, semi-
structured and unstructured formats, and arrive
at an ever-increasing velocity and complexity.
Analyzing this data is now mission-critical, since
it can provide more timely and precise insights
to guide business planning and decision-making.
With so much transparent content generated
daily through social media, data science can
help banks deliver a consistent and integrated
customer experience.
To use this data for business advantage, banks
must set up data analysis teams to collect, sift and
apply meaning from this data to advance business
goals. According to Gartner, big data in the
banking industry has the highest level of oppor-
tunity because of the high volume and velocity of
data in play. Moreover, 78% of CFOs have labeled
BI and analytics as the top technology initiative
for their departments — beating out even financial
management applications.3
Key Inputs for Data Science
As noted earlier, data can be broadly categorized
as structured and unstructured. At a broad level,
structured data comprises transactional data,
which includes customer buying/spending habits,
and unstructured data can be obtained from
various social media sites, such as Facebook and
Twitter. Precise analysis of social data is of great
importance because it provides valuable insight
into individual customers’ likes, dislikes, prefer-
ences, etc.
Analysis of both structured and unstructured data
can help banks better target the right product to
the right customer at the right time. For example,
by correlating the social activities (unstructured
data) of a customer with a spending pattern
(structured data), banks can customize and
optimize the timing of their product offerings.
For even more precise targeting, organizations
can add new third-party data sources, compiled
from a variety of sources, such as public reposi-
tories, mobile devices and cars. As such, data
science involves three aspects of data: velocity,
volume and diversity (see Figure 1).
cognizant 20-20 insights

3cognizant 20-20 insights
Data Science: Usage Areas
Many business areas can benefit from data sci-
ence (see Figure 2). To properly ascertain how
customers prefer to be served, banks can apply
such data science techniques as hypothesis test-
ing, crowdsourcing, data fusion and integration,
machine learning, natural language processing,
signal processing, simulation, time series analysis
and visualization. Using the insights gleaned from
these approaches, marketers can derive the right
marketing strategy through a mix of marketing
messages and offers that resonate with individual
customers and segments.
For example, using a mobile app, banks can
analyze individual consumer behaviors and
spending activities and combine that data with
credit bureau information. When analyzed, the
resulting insights can lead to better targeted
messaging around a potential offer, such as a
pre-approved home loan to a customer who is
qualified based on analysis of the data contained
in his transactional files and interactions on social
media.
The vast amounts of online data have much
to offer banks seeking consumer insights. For
instance, by combining information from travel
Web sites and spending patterns gleaned from
internal databases, banks can optimize their
product mix and offers. Analysis of transactional
behavior like recency, frequency and monetary
value can be sliced and diced to derive customer
profiles that can improve the effectiveness and
efficiency of targeted marketing efforts. An
example is an Australian bank that is working
with a retailer to better understand where the
retailers’ customers live, when and where they
shop, and how much they spend. This informa-
tion is then used to refine the retailer’s branch
location/relocation strategy.4
Another example is a bank that uses point of sale
data to determine whether a customer frequents
a certain area for shopping or lunch and then use
this information to deliver online offers that are
highly personalized even to the type of food the
customer prefers, increasing the probability that
the offer would be accepted. Adding device-spe-
cific capabilities, the offer could be delivered by
SMS at the most logical time for decision-making.
Data Science Trio
•
Velocity
• Batch process
• Near real-time
• Real-time
Volume
• Records in terabytes,
petabytes
Diversity
• Structure transactional data
• Unstructured/semi-structured data
from social source
Figure 1
Applying Data Science
Intelligent
Forecasting
Consumer
Sentiment
Fraud
Detection
Customer
Service
Target
Marketing
Data
Science Areas
of Usage
Consumer
Proﬁling
Figure 2

Quick Take
As an early warning system, data science solutions
can help banks quickly identify potentially
fraudulent behavior before the fraud becomes
material. For example, individual cardholders are
creatures of habit. Cardholders have “favorites“
or recurrences over a wide variety of objects in
their transaction streams. These objects might
include favorite ATMs that are close to work or
home or gas stations along a daily commute, as
well as preferred grocery stores and online sites
for shopping.
An analytics technique that could be used to
improve fraud management is to identify card-
holder favorites, in order to distinguish between
“in-pattern,” or normal, customer spending and
“out-of-pattern” suspicious transaction activity.
This enables faster fraud detection at much lower
false positive rates (declines on legitimate trans-
actions).
Text analytics of unstructured data can help banks
identify patterns of information that indicate the
likelihood of fraud. Text mining of insurance claim
descriptions (written and recorded) provided by
bogus claimants uncovered some very interest-
ing facts. It turns out that certain phraseologies
(the use of “ed” rather than “ing” on the end of
verbs, for instance), are extremely indicative of
fraudulent claims. This is due to the different ways
in which people relay stories they actually expe-
rienced vs. those they concocted; for instance “I
was walking” is indicative of someone recounting
an actual experience whereas “I walked” often
turns out to be indicative of someone describing
a fictitious event.
Applying Data Science to Detect Fraud Before The Fact
4cognizant 20-20 insights
This is the same approach perfected by Amazon
and other retailers.
Unstructured data, such as social media com-
ments, can help banks gain insight into what cus-
tomers like and don’t like about various brands,
products and service and also gather feedback
about their own products and services. By closely
tracking customer comments, banks can quickly
identify issues and take action to improve the cus-
tomer experience. The instant feedback of social
media also enables banks to capitalize on oppor-
tunities to proactively counteract negative per-
ceptions, exceeding customer expectations and
driving loyalty. Banks can also use social media
data to target customers with offers or services
aligned with recent life events (e.g., graduation,
marriage, new job).
Data science can help banks recognize behavior
patterns, providing a complete view of individual
customers and segments. For example, when a
customer enters a bank, customer representa-
tives can be better equipped to offer the right
products and provide a quicker resolution to
customer queries by analyzing their Code Halos.
Data science can also be used by banks to analyze
the average cost for each channel (e.g., call center,
branch banking, etc.) and design strategies to
migrate customers to low-cost channels.
Analytics techniques can also play a signifi-
cant role in the early warning, detection and
monitoring of fraud. These techniques allow
organizations to extract, analyze, interpret and
transform business data to help detect potential
instances of fraud and implement effective fraud
monitoring programs (see sidebar).
Advanced data science techniques could enable
institutions to improve underwriting decisions
and increase revenues while reducing risk costs.
These techniques can be fruitful across all asset
classes, all types of credit risk models and the
entire credit life cycle, including profit maximiza-
tion and portfolio management.
For debt collections and recoveries, analytics
is a critical part of the process, as it can enable
organizations to create an accurate picture of
the customer’s propensity and ability to pay and,
therefore, the amount likely to be recovered. This
behavioral scoring is used to segment customers
and prioritize collections activities to maximize
recoveries and reduce collections costs.
Overcoming Challenges
What follows are the common obstacles banks
encounter when attempting to implement an
effective data science strategy.

cognizant 20-20 insights 5
Data Volume
Over the last decade, banks have accumulated
huge volumes of data, especially following the
introduction of smartphones, tablets and now
wearables that enable multi-channel access;
however, many still suffer from a scarcity of
insight. Managing enormous data sets, as well as
analyzing and correlating structured, semi-struc-
tured and unstructured formats, makes the data
science job increasingly complex.
Distinguishing “signal” (meaningful insight)
from “noise” (massive amounts of unmanaged
data) remains a fundamental challenge and a
significant opportunity. There are various data
cleansing techniques, such as clustering, outlier
detection, etc., that can help organizations find
correlations within date sets.
Budget Constraints
Banks must be willing to invest significantly in
people, infrastructure and platforms to effective-
ly analyze and make strategic decisions from big
data. Beyond these investments, such initiatives
also need to strategically align with the bank’s
overall vision and business mission. Such initia-
tives require qualitative and quantitative scrutiny
in order to prioritize the projects with the highest
payback. Priorities can be determined by strategic
and tactical benefits, cost, duration, people and
technology availability.
Privacy Concerns
Gaining permission to use and process data from
mobile and social media is a huge challenge.
Numerous concerns have been raised over
identity theft, privacy and social media stalking,
among other issues. Within the bank, it is also
important to ensure that the right people across
the organization (i.e., bank decision-makers) can
access the right data, at the right time.
Organizations must also decide who owns the
data before a data science project is implement-
ed, so that accountability and workflow can be
properly set and followed.
Skilled Talent
There is a huge demand for data scientists, and
the pool of available talent is insufficient to meet
the needs of every organization. Finding highly
skilled data scientists is not easy; they do not
simply report on data but look at it from many
angles, running complex queries to find correla-
tions and patterns. They also need to communi-
cate their findings and recommendations to top
leadership. Some of the top skills required for
data scientists include analytics know-how, statis-
tical acumen, domain expertise data mining and
the ability to clearly and effectively communicate.
Looking Forward
Today’s knowledge economy provides businesses
of all kinds with access to big data that’s growing
exponentially in volume, variety, velocity and
complexity. With more data coming from more
sources faster than ever, the questions will only
continue to unfold. Some examples:
• What is your organization’s data science
strategy?
• How is your enterprise combining new and
existing data sources to make better decisions?
• How could new data sources, including social,
sensors, location and video, help improve your
organization’s business performance?
• Will your organization take advantage of big
data or remain paralyzed through endless
analysis?
A savvy, experienced team of data science con-
sultants can help organizations create a roadmap
that results in a meaningful, business-aligned
approach to data science. Experts can help
implement data science technologies, manage big
data, accurately predict customer demand and
make better decisions faster than ever before.
The best approach is to start small rather than
setting off a big bang. The mantra for successful
data science projects depends on the organiza-
tion’s business objectives, but one constant is
focus and agility. For example, if the business
need is to define customer segments to drive
pricing-elasticity models, the IT organization
should first discover which customer data needs
to be gathered before building an enterprise data
warehouse and an enterprise analytics platform.
Experts can develop an initial proof of concept by
analyzing the internal, external, structured and
unstructured data and conclude with meaningful,
business-aligned recommendations.
A blueprint can help guide the organization to
develop and implement data science solutions in
ways that deliver business value. From there, an
implementation strategy followed by a detailed
plan can be built (see Figure 3, next page).

cognizant 20-20 insights 6
Data Science Implementation Plan
Figure 3
3
BussiinneessB
oobjective
Internal, external,I
social data analysis by
data science
ccoonnssuullttaanntt
cept note/Concept note
proof off ccoonnceptpro
ImplementatiioonnImplementatio
ssttrraatteegy
ailedDetailed
iimmppllementatioonn
plan
About the Authors
Shantanu Dubey is a Consultant within Cognizant Business Consulting’s Banking and Financial Services
Practice. He has over seven years of business and IT consulting experience in implementation of BASEL,
regulatory reporting, business intelligence and core banking solutions. Shantanu also has experience
working with leading banks on product development, business process optimization, business require-
ments management and gap analysis across various geographic locations. He holds a bachelor’s
degree in information and technology engineering from RGPV Bhopal, and a post-graduate diploma in
management from I2it Pune. Shantanu can be reached at Shantanu.dubey@cognizant.com.
Siddhartha Nainwani is a Consultant within Cognizant Business Consulting’s Banking and Financial
Services Practice. He has over seven years of business and IT consulting experience, working with
leading banks in business process management, business analysis and test management across various
geographic locations. Siddhartha holds a bachelor’s degree in engineering in information technology
from Shivaji University, Maharashtra, and a master’s degree in management from ICFAI Business School,
Mumbai. He can be reached at Siddhartha.Nainwani@cognizant.com.
Footnotes
1
For more information on Code Halos, please see our white paper, “Code Rules: A Playbook for Managing
at the Crossroads,” or our recently published book, Code Halos: How the Digital Lives of People, Things
and Organizations Are Changing the Rules of Business.
2
Wikipedia definition, http://en.wikipedia.org/wiki/Big_data.
3
“Three Reasons Why BI and Analytics Is The Top CFO Initiative,” Domo,
http://www.domo.com/learn/3-reasons-why-bi-analytics-is-the-top-cfo-initiative.
4
Anthony Duffy, “Unlocking the Potential of Big Data,” Banking Technology,
http://www.bankingtech.com/58812/unlocking-the-potential-of-big-data/.

About Cognizant
Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process out-
sourcing services, dedicated to helping the world’s leading companies build stronger businesses. Headquartered in
Teaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep industry
and business process expertise, and a global, collaborative workforce that embodies the future of work. With over 75
development and delivery centers worldwide and approximately 178,600 employees as of March 31, 2014, Cognizant
is a member of the NASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked among
the top performing and fastest growing companies in the world. Visit us online at www.cognizant.com or follow us on
Twitter: Cognizant.
World Headquarters
500 Frank W. Burr Blvd.
Teaneck, NJ 07666 USA
Phone: +1 201 801 0233
Fax: +1 201 801 0243
Toll Free: +1 888 937 3277
Email: inquiry@cognizant.com
European Headquarters
1 Kingdom Street
Paddington Central
London W2 6BD
Phone: +44 (0) 20 7297 7600
Fax: +44 (0) 20 7121 0102
Email: infouk@cognizant.com
India Operations Headquarters
#5/535, Old Mahabalipuram Road
Okkiyam Pettai, Thoraipakkam
Chennai, 600 096 India
Phone: +91 (0) 44 4209 6000
Fax: +91 (0) 44 4209 6060
Email: inquiryindia@cognizant.com
© Copyright 2014, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is
subject to change without notice. All other trademarks mentioned herein are the property of their respective owners.

Bank(ing) on Data Science

More Related Content

What's hot (19)

Similar to Bank(ing) on Data Science (20)

More from Cognizant (20)

Recently uploaded (20)

Bank(ing) on Data Science