Using Advanced Analytics to Combat P&C Claims Fraud

thumbsshoesSoftware and s/w Development

Dec 11, 2013 (3 years and 10 months ago)

75 views


leadership.
-
1
fraud, not to mention the resulting losses. The
these vast data pools is to pursue analytics as a
vert it into insights that quickly aid in identifica
which demands a major commitment from senior



tion and avoidance of fraudulent claims. By using
ships among various parties involved in a claim

insurers to identify common types of fraud early
in silos, resulting in data inconsistencies. One
requires P&C insurers to embrace predictive and advanced analytics, such as
|
analytics, insurers can establish relation
detect highly sophisticated fraud, as well as
ers deal with an uncertain economic climate and
an efficient model and approach to enterprise-
systematic use of advanced analytics that detect
-
can deliver analytics as a service, insurers can improve their bottom lines,
the National Insurance Crime Bureau.
tinuous flow of quality data across various func
text, social media, link and geospatial analysis. By partnering with firms that
in the claims process. Advanced analytics, such
help insurers sift through and draw inferences
wide data management. Insurers must focus on
reduce fraud-related losses, as well as condense
crime in the U.S. after tax evasion, according to
organized crime rings.
tional areas of the organization to enable a more
Cognizant Reports
december 2012
breaking down data silos and ensuring a con
and prevent fraud. Getting there requires a
-
from unstructured data more quickly and con
As insur
Combating the growing complexity and sophistication of claims fraud
Insurance fraud is the second biggest white-collar

the claims cycle, resulting in improved customer
satisfaction. Historical claims data, combined
-
based on geographic location. This can help
with industry data, can be a starting point for
Achieving this level of sophistication requires
cognizant reports
as social media analytics and text mining, can
enhance claims processing efficiencies and boost customer satisfaction.
Using Advanced Analytics to Combat

P&C Claims Fraud
link analytics in combination with geospatial
traditional methods of identifying fraud are no
Advanced analytics can help insurers identify and

cultural shift toward fact-based decision-making,
way to quickly and effectively extract value from
on traditional database systems that operate
the increasing incidence and sophistication of
intense competition, they must also grapple with
-
service (AaaS), a new delivery model that enables

-
longer sufficient.
Executive Summary
However, many insurers still run their business
cognizant reports
2
insurers to work closely with specialists who

provide analytical insights on a pay-per-use

basis. This model shifts the cost of owning tech
-
nology infrastructure, processes and talent to the
chosen partner.
Fraud: A Growing Menace
On average, insurers lose $30 billion annually
to fraudulent claims, representing 10% of their
claims expenses, according to the Insurance Infor
-
mation Institute (see Figure 1).
2
Insurance fraud
can be divided into two categories: opportunistic/
soft fraud and professional/hard fraud. Oppor
-
tunistic fraud is committed by individuals who
inflate damages or repairs in a legitimate claim or
provide false information to reduce the premium
amount. About half of P&C insurers lose 11 cents
to 30 cents or more per premium dollar to soft
fraud alone, according to the Insurance Research
Council-Insurance Services Office.
3
Professional, or hard, types of fraud are

committed by organized groups that steal

vehicles, deliberately damage property and stage
accidents. These gangs are well acquainted with
fraud detection systems and collude with doctors,
attorneys, insiders in insurance companies, body
shops, etc. to lodge fraudulent claims.
A growing concern for insurers is the increas
-
ing number of questionable claims
4
referred to
the National Insurance Crime Bureau (NICB) by
its member insurance companies. Between 2010
and 2011, property-related questionable claims
increased by 2%, while casualty-related claims
Estimated Annual Loss* Due to Fraud
Figure 1
*Assuming 10 of P&C claim expense
Source: Insurance Information Institute, July 2012
($B)
27.2
28.0
28.7
30.1
31.2
28.7
30.2
34.3
30.9
31.3
34.8
0
5
10
15
20
25
30
35
40
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
increased by 16% (see Figure 2, next page).

The prolonged weak economy is inflicting

significant economic hardships for consumers
and businesses, which is further increasing the
cost of fraud, especially in personal lines, accord
-
ing to 54% of the 143 insurers surveyed in August
2012 by FICO and the Property Casualty Insurers
Association of America (PCI).
5
Fraud negatively impacts insurers’ bottom lines
(reduced profitability due to the cost of fraudulent
claims that would otherwise not be incurred) and
competitiveness (delays in claims processing). It
increases premiums for customers, as insurers
charge them more to make up for the increase
in payouts. NICB estimates that fraud increases
premiums by $200 to $300 per family, annually.
The Need for Analytics
The P&C insurance industry continues to

operate in an uncertain economic climate, with
low interest rates hampering investment income.
The direct loss ratio rose by 2.7 points from
2010, to 67.5% in 2011,
6
while the combined ratio
in the first half of 2012 was 102%.
7
Fraud, along
with long-tail liabilities such as incurred but not
reported (IBNR) liabilities, produce uncertainty,
making it much tougher to assess accurate claims
reserves and pricing of premiums.
Compliance with the Dodd-Frank Wall Street
Reform and Consumer Protection Act
8
and

the expected impact of Solvency II
9
beyond

EU borders requires U.S. insurers to invest
in enterprise risk management and related

cognizant reports
3
support systems, adding to already strained

operating costs. There has also been an increase
in the frequency of natural catastrophes (see

Figure 3, next page) and, consequently, in the
cost of serving customers. Superstorm Sandy
cost insurers between $20 billion and $25 billion,
according to disaster-modeling company Risk
Management Solutions Inc.
10
Insurers are there
-
fore looking to significantly reduce costs and
improve process efficiencies.
Claims are at the heart of P&C insurer

operations and account for about 80% of their
costs. An efficient claims service is crucial for
creating a sustainable customer relationship.

Further, with long-tail liabilities looming, timely
management of claims becomes very important.
However, the claims departments at many

insurers are hamstrung by outdated tools and
a shortage of qualified employees. Claims staff
at many organizations must devote at least half
of their time to routine administrative tasks.
Identifying fraudulent claims early improves
claims processing. It reduces cycle time as

suspicious claims are weeded out and sent for
further investigation, while legitimate claims
are prioritized. This, in turn, results in improved

customer service, as well as significant savings
for the organization.
Insurers have always had systems in place to
identify fraudulent claims and special teams
to investigate suspicious claims. However, the

growing complexity of fraud and well-executed
fraud schemes have exposed the limitations
of traditional fraud-detection systems, such as

internal audits, whistleblower hotlines to report
fraud and software that flags anomalies based on
a pre-defined set of rules.
Questionable Property Claims
Questionable Casualty Claims
Figure 2
0
2,000
4,000
6,000
8,000
Flood/water Suspicious
disappearance/
loss of jewelry
Inflated
damage
Suspicious
theft/loss
(excluding vehicles)
Fire/arson Hail damage
2009
2010
2011
2009
2010
2011
Source: NICB, February 2012
0
4,000
8,000
12,000
16,000
20,000
Duplicate
billing
Excessive
treatment
Billing for
services not
rendered
Unbundling/
upcoding
Faked/
exaggerated
injury
Prior
injuries
Jump-in
Inflated
billing
Slip and fall
Solicitation
Provider/facility
improperly licensed
/incorporated
Staged/caused
accident
cognizant reports
4
Such systems can detect only some types of fraud
(usually soft fraud) and not early enough for

preventive action. They have also been known
to cause major embarrassment by flagging
legitimate claims as fraudulent. Additionally, in
a bid to retain customers in a highly competitive

environment, insurers have refrained from

making a serious effort to investigate suspi
-
cious cases. Not surprisingly, fewer than 20% of

fraudulent claims are detected.
11
Insurers have large amounts of data that can
help identify fraud. However, the data is usually

scattered across organiza
-
tional silos and exists as
unstructured data, making
it practically impossible to
use it for gaining meaning
-
ful insights. To deal with
this, insurers must adopt
an enterprise-wide data-
centric approach, clean and
integrate the historic claims
data collected over the
years and stored in dispa
-
rate databases, and develop
predictive models to gain a

complete view of custom
-
ers and their transactions. This can help identify
a variety of fraud types quickly and effectively,
reducing losses significantly.
Catastrophe Insured Losses
Figure 3
* Only Sandy-related losses, as estimated by Risk Management Solutions, Inc.
U.S. CAT losses in 2011 were the fifth highest in U.S. history on an inflation adjusted basis.
Note: 2001 figure includes $20.3B for 9/11 loss reported through 12/31/01. Includes only business and personal
property claims, business interruption and auto claims. Non-property/BI losses = $12.2B ($15.6B in 2011 dollars).
Source: Property Claims Service/ISO and Insurance Information Institute
13.7
4.7
7.8
36.9
8.6
25.8
12.3
10.7
3.7
14
11.3
6
33.9
7.4
15.9
32.9
71.7
10.3
7.3
28.5
11.2
14.1
32.6
25*
0
20
40
60
80
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
($B)
Predictive analytics
helps to quickly and
more accurately
determine
whether a claim
needs further
investigation

and to determine

the complexity of
the claim.
Analytics for Improved Fraud Detection
Insurers generate large volumes of customer
data, such as policy details, previous claims

and information gathered from adjusters. This
data can be used in combination with data from
industry sources such as NICB to run predictive
analytics to identify fraudulent claims early in the
claims process.
For example, NICB’s ForeWARN database allows
member companies to search and identify
whether a party had committed fraud in the past
and obtain additional information to develop
fraudulent patterns and trends. NICB also pro
-
vides analytics support to member companies
to identify fraud patterns and exposure, helping
organizations in fraud investigations.
12
Predictive analytics, which involves the use of
regression models and advanced techniques such
as neural networks, helps to quickly and more
accurately determine whether a claim needs fur
-
ther investigation and to determine the complex
-
ity of the claim. This speeds up the processing of
legitimate claims, resulting in improved customer
satisfaction, as well as preventing payouts for
fraudulent claims. It also aids in assigning staff
with the appropriate level of experience based on
the severity of a claim. Insurers have also begun
deploying expert systems that apply artificial
intelligence algorithms to proactively identify
fraudulent activities.
cognizant reports
5
In addition, by combining social network and
social media analytics, link analysis and geospa
-
tial analysis, insurers can identify fraud that is
hard to detect using traditional methods.
Social Network and Social Media Analytics
Customers share varying degrees of relationships
with other individuals with whom they share
group membership. Social network analytics,

for example, helps to identify proximities and

relationships among people, groups, organiza
-
tions and related systems. It reveals the strength
of the relationships and how information flows
within the groups and, most importantly, group
influencers. This provides valuable input on
whether a customer is affiliated with any fraudu
-
lent group and helps to predict the chances of a
particular customer committing fraud.
Insurance companies are also increasingly min
-
ing social media to detect and investigate fraud.
With two out of three people in the U.S. using
social networking sites, tracking customers’
social media updates can help insurers investi
-
gate suspicious claims. By tracking social media
accounts and applying social network analytics
to the information on social media, insurers can
gain information about claimants, medical provid
-
ers, body shops, etc., as well as a claimant’s con
-
nection with organized crime networks. Investiga
-
tors in California recently used Facebook to find
that four women, who staged an auto accident to
defraud insurers of about $40,000 and denied
knowing each other, were in fact friends.
A majority of social media users are either

ignorant of the security settings that hide their
information from others or do not bother to
enable them, thus providing claims’ investigators
with clues. For instance, Facebook’s location ser
-
vice, which allows users to update their locations,
offers investigators insights into whether a car
driver visited a bar before hitting a tree.
Text Mining
Text mining and predictive modeling will be the
primary tools that insurers will deploy to com
-
bat fraud in the next two years, according to a
2012 study by SAS Institute and Coalition Against
Insurance Fraud (CAIF) of 74 U.S. insurance exec
-
utives.
13
Text analytics helps companies gain critical
insights from large volumes of unstructured data,
such as adjuster notes, first notice of loss, e-mail
and accident descriptions, which usually consist
of short or incomplete sentences, misspelled
words and abbreviations. Based on the key words
used to describe an incident, text analytics helps
insurers detect attempted fraud by flagging
questionable incidents, exaggerated injuries and
treatment costs, reckless driving, etc. and recom
-
mends actions.
For example, an adjuster’s notes of an injured
customer that contains key phrases such as “car
moving slowly,” “head-on collision with another
slow-moving car,” “complains of severe neck
pain,” “reports excessive treatment costs,” etc.
can help insurers determine whether the claim
needs to be probed further. This ensures that only
cases with strong fraud patterns are forwarded
to investigation units and improves an adjuster’s
ability to quickly process genuine claims.
Link Analysis and Geospatial Analysis
An individual claim may not appear false at first
glance. Often, it is only when it is seen in the

context of previous fraudulent claims, or claims
with a high fraud score, that those anomalies
become apparent. Link analysis provides that
larger picture for a claim. In the case of a car
accident involving multiple claimants, link analy
-
sis can use claimants’ addresses, phone numbers,
vehicle numbers, etc. to unearth links among the
claimants, the clinics where the claimants were
treated and the body shop they used, thus leading
investigators to rings of professional fraudsters.
While link analysis allows investigators to under
-
stand whether the parties involved in a large
group of injury claims are interrelated, geospatial
analysis can provide location-based information
related to a claim, as well as the physical prox
-
imity of the claimants and others involved in a
claim. In the case of a staged accident, geospatial

analysis provides information about the loca
-
tion of the accident, the distance between the

various claimants’ residences and their proximity
to resources such as a lawyer, a body shop and

a medical provider. This provides investigators
with evidence to pursue a hunch and to identify
potential fraud rings.
Geospatial analysis can also be used to identify
the exact area affected by a natural disaster or
an explosion, which helps determine the amount
of risk to insured properties and weed out claims
that are filed from areas that are not located in
the affected zone.
cognizant reports
6
Challenges
Insurers generally use a combination of anti-
fraud technologies. Older technologies, such as
red flags/business rules and scoring capability,
still dominate the scene, according to the CAIF
and SAS survey. Fewer than 50% of respondents
use more advanced techniques, such as workflow
routing, text mining, predictive modeling and
geographic data mapping, while 12% do not use
any anti-fraud technologies, the survey found.
A major obstacle to embracing analytics is the
lack of enterprise-wide data management at
many insurers. While insurance companies are
data-rich, not many have made progress on
the data management front. Much of their data
resides in numerous independent legacy systems,
often resulting in data inconsistency. It is, there
-
fore, important that data structures across the
organization be standardized and inconsistencies
resolved to realize the full benefits of analytics.
Other major challenges in deploying analytics
include the lack of IT resources and concerns
about return on investment, according to the CAIF
and SAS study (see Figure 4).
14
Some insurers also
cite legal and compliance issues that can arise
from using social media data for investigations.
Overcoming Obstacles
To leverage the benefits of advanced analytics,
insurers need to focus on fresh approaches to
data management that can integrate disparate
systems and effectively deal with data overflow.
By integrating predictive analytics with enterprise
systems, insurers can build real-time analytical

capabilities that help in creating a just-in-time
understanding of opera
-
tional issues, effective
fraud identification and
more meaningful and
timely decisions. A large
U.S. insurance company
that deployed real-time
analytics to sift through
unstructured claims data
from two fraud-prone states
found that more than 1,000
insured customers were
actually high-risk custom
-
ers. Another insurer identi
-
fied actionable claims worth
$20 million within the first
three months of deploying
fraud analytics.
15
Benefits
While there is no denying that deployment of
advanced analytics requires significant invest
-
ment, the benefits far outweigh the costs. Some
examples:

Efficient fraud detection reduces annual claims
payouts.

The number of false positives identified and
pursued is minimized. This boosts employee
productivity, minimizes loss adjustment
expenses and avoides customer ire and legal
hassles. Advanced analytics helped a U.S.
insurer improve its false-positive detection
rate by 17%.
16
Challenges in Deploying Analytics
36
38
14
7
5
Cost/benefit analysis (ROI)
Lack of IT resources
Proof of concept and unknown effectiveness
Acquisition and integration of data
Legal and compliance issues
Figure 4
Source: The State of Insurance Fraud Technology, Coalition Against Insurance Fraud and SAS, September 2012
A large U.S.
insurance company
that deployed

real-time analytics
to sift through
unstructured claims
data from two
fraud-prone states
found that more
than 1,000 insured
customers were
actually high-risk
customers.
cognizant reports
7

Claims processing cycle time can be reduced,
resulting in faster processing of claims and
increased customer satisfaction.

Losses through payouts can be minimized,
thus eliminating the need to increase premi
-
ums and thereby helping to build strong cus
-
tomer relationships. Santam, a South African

short-term insurer, saved $2.4 million on fraud
-
ulent claims within four months of deploying
analytics. It also improved fraud detection
capabilities and unearthed a motor fraud ring
within one month of deploying analytics.
17

Insurers committed to fighting fraud will be
able to send a strong message to fraudsters

and enhance their image in the eyes of

customers.
Embracing Analytics as a Service
The growing complexity of fraud requires

organizations to move beyond rules-based and
judgmental approaches toward more fact-based
and self-learning analytical models. We believe an
ideal fraud detection approach must combine the
best of analytics and rules-based approaches.
Insurers acknowledge that deploying predictive
analytics is the most effective way to combat fraud,
according to the FICO and PCI study. The increas
-
ing confidence in analytics is reflected in the rise in
data and analytics budgets. According to a recent

survey by Strategy Meets Action of 165 insur
-
ers, three quarters of the respondents plan to
increase their annual data and analytics spend
-
ing between 2012 and 2014, with 19% planning
to increase outlays by
more than 10% per year

(see Figure 5).
Major insurers have
employed statisticians and
predictive modelers and are
capable of building efficient
fraud detection models.
However, many insurers
are revisiting their decision
to build in-house capabili
-
ties due to the complexities
involved in handling analytics and the expertise
required for text mining, using social media and
geospatial analysis. While in-house solutions offer
greater control over development, “operational
-
izing” a fraud detection model and the infrastruc
-
ture required to implement and run an analytical
solution can be expensive.
18
Vendor solutions,
on the other hand, offer lower total cost of

ownership.
Open source projects, such as R and Apache
Hadoop, are being used by organizations to do
more with big data. While Apache Hadoop helps
to efficiently store and manage huge volumes
of data, R is widely used for data manipula
-
tion, calculation and graphical display. Further,
by combining R and Hadoop, organizations can

overcome the complexity of processing large

volumes of unstructured data and analyzing
social media networks in short periods.
Insurers' IT Spending Plans for Data and Analytics (2012-14)
Figure 5
n=165
Source: SMA Research, Data and Analysis, 2012
19
21
35
23
2
Increase by more than 10 per year
Increase by 6-10 per year
Increase by 1-5 per year
Spending will remain flat
Decrease
Insurers
acknowledge
that deploying
predictive analytics
is the most
effective way to
combat fraud,
according to the
FICO and PCI study.
cognizant reports
8
However, deploying analytics is no easy
task. Unstructured data accounts for about
80%
19
of organizational data and is bound
to grow at 60% annually,
20
with the increas
-
ing chatter created on social media. The

traditional IT infrastructure deployed by most
insurers is insufficient to analyze such large

volumes of data and requires organizations
to invest in people, processes, IT tools and

infrastructure.
Choosing the Right Partner
With process virtualization and cloud comput
-
ing, opportunities now exist for cost-cutting
through global sourcing via
the business process as a

service (BPaaS) model. This
can save precious Cap-Ex
by transferring the cost of
acquiring expensive hard
-
ware, software and key talent
through consumption-based
pricing models.
A subset of BPaaS, analytics
as a service combines tradi
-
tional knowledge process out
-
sourcing (KPO) and business
process outsourcing (BPO)

capabilities with more effi
-
cient, cloud-enabled ways
of delivering analytical insights. This approach
allows organizations to deploy analytics solu
-
tions tailored to their needs. The service can be
increased or decreased as business requirements
dictate, providing more Op-Ex flexibility.
Organizations should seek a partner that can
seamlessly marry analytics with technology
rather than a pure-play analytics player that
may not have industry domain expertise. The key

analytical component is derived from the

ability to understand various forms of insurance
fraud — ranging from early payment defaults to
more complex types of malfeasance — and devel
-
oping predictive models capable of understand
-
ing complex relationships and learning from his
-
torical data.
The partner must have expertise in

extracting meaningful insights from insurance-
related social networks and
social media and perform
complex analyses on the
data. The technology com
-
ponent includes the part
-
ner’s ability to integrate
advanced analytics with
insurers’ claims systems,
and create new claims effi
-
ciencies and improve over
-
all claims effectiveness.
As analytics processes
become standardized and
can uniformly be applied
via cloud-enabled models
(harnessing the growing clout of utility comput
-
ing architectures), we believe that insurers stand
to benefit greatly by associating themselves with
partners that have invested in such capabilities.
Looking Forward
To experience the potential of analytics, we
believe that insurers should:

Develop an enterprise-wide data architecture.

Identify key areas for deploying analytics.

Design a comprehensive strategy for adoption
and implementation of analytics, including
information technology.

Develop a fact-based decision-making culture
focused on achieving specific goals.

Formulate customized strategies to capitalize
on unique data.

Continuously innovate and renew analytics
implementation.

Enter into relationships with partners capable
of providing AaaS to advance competitive
advantage.
The traditional
IT infrastructure
deployed by
most insurers
is insufficient
to analyze large
volumes of data
and requires
organizations to
invest in people,
processes, IT tools
and infrastructure.
Organizations
should seek a
partner that can
seamlessly marry
analytics with
technology rather
than a pure-play
analytics player
that may not have
industry domain
expertise.
cognizant reports
9
Footnotes
1
“Insurance Fraud: Understanding The Basics,” NICB, April 21, 2011,
https://www.nicb.org/File%

20Library/Theft%20and%20Fraud%20Prevention/Fact%20Sheets/Public/insurancefraudpublic.pdf
.
2
“Insurance Fraud,” Insurance Information Institute, June 2012,
http://www.iii.org/assets/docs/pdf/
InsuranceFraud-072112.pdf
.
3
“Fraud Stats,” Coalition Against Insurance Fraud,
http://www.insurancefraud.org/statistics.htm
.
4
According to NICB, a questionable claim is one that NICB member insurance companies refer to NICB
for closer review and investigation based upon one or more indicators of possible fraud. A single

questionable claim can contain up to seven different referral reasons.
5
“FICO PCI Insurance Fraud Survey,” FICO, October 4, 2012,
http://www.fico.com/en/Company/News/
Pages/10-04-2012.aspx
.
6
“Written Premium, Rising Loss Ratios Point to Continued Rate Increases,” PropertyCasualty360,
March 27, 2012,
http://www.propertycasualty360.com/2012/03/27/written-premium-rising-loss-ratios-
point-to-contin
.
7
“Property Casualty Insurers Benefit From Drop In Catastrophe Losses,” Property Casualty Insurers
Association of America, October 4, 2012,
http://www.pciaa.net/LegTrack/web/NAIIPublications.nsf/

lookupwebcontent/4003FCCDBDDDB35186257A8E006C244?opendocument
.
8
The Dodd-Frank Wall Street Reform and Consumer Protection Act requires insurers to comply

with data requests on sales, customer location, etc. from the Federal Insurance Office (FIO), Office of
Financial Research (OFR), in addition to state regulators. Further, insurers with more than $50 billion
in assets are identified as systematically important financial institutions (SIFI) and have to comply
with heightened regulations. This means big insurers must rewire their legacy systems to create more
effective reporting mechanisms, which is expensive and can result in a competitive disadvantage
when compared with smaller insurance companies.
9
Scheduled to come into effect on January 1, 2014, Solvency II requires U.S. insurers operating in

the European Union to comply with the regime’s new risk management and reporting and account
-
ing standards. Meeting the data requirements and reporting frequency of Solvency II requires U.S.

insurers to invest in revamping and consolidating existing IT systems.
10
“Sandy May Cost Insurers Up To $25 Billion,”
The Wall Street Journal
, November 14, 2012,
http://online.
wsj.com/article/SB10001424127887324735104578119301366617508.html
.
11
“Predictive Analytics: A Powerful Weapon In Fight Against Fraud,” PropertyCasualty360, April 4, 2011,
http://www.propertycasualty360.com/2011/04/04/predictive-analytics-a-powerful-weapon-in-fight-ag
.
12
“Join NICB,” NICB,
https://www.nicb.org/about-nicb/join_nicb
.
13
“The State of Insurance Fraud Technology,” SAS Institute, September 2012,
http://www.sas.com/reg/
wp/corp/50373
.
14
“The State of Insurance Fraud Technology,” Coalition Against Insurance Fraud, September 2012,

http://www.insurancefraud.org/downloads/techStudy_2012.pdf
.
15
“Predictive Analytics Can End the Isolation,” PropertyCasualty360, October 1, 2012,
http://www.

propertycasualty360.com/2012/10/01/predictive-analytics-can-end-the-isolation
.
16
Ibid.
17
“Using IBM Analytics, Santam Saves $2.4 Million in Fraudulent Claims,” IBM, May 9, 2012,
http://

www-03.ibm.com/press/us/en/pressrelease/37653.wss
.
cognizant reports
10
References

“Uncertainty Weighs Down U.S. Insurers,” Life Insurance International, July 2012,
http://www.

goldbergsegalla.com/sites/default/files/uploads/JW_LifeInsuranceInternational_July2012.pdf
.

“You Know There Are Fraudulent Claims. Let’s Find Them Now,” Statsoft, 2011,
http://www.statsoft.
com/portals/0/solutions/StatSoft_InsuranceFraud_Brochurev.pdf
.

“The Three Best Targets For Attacking P&C Insurance Fraud,” FICO, September 2010,
http://www.efma.
com/efmaweb_files/file/Partnerships/Fico_Insights44_Insurance_Fraud.pdf
.

“Driving Operational Excellence in Claims Management,” February 2011,
http://www.deloitte.
com/assets/Dcom-UnitedStates/Local%20Assets/Documents/FSI/US_FSI_DrivingOperational

ExcellenceInClaimsManagement_022311.pdf
.

“Advanced ‘Big Data’ Analytics with R and Hadoop,” Revolution Analytics, 2011,
http://www.revolution
-
analytics.com/why-revolution-r/whitepapers/R-and-Hadoop-Big-Data-Analytics.pdf
.

“Identify Claims Fraud with Advanced Analytics: Uncover Fraud that Traditional Methods Cannot Find,”
FICO, November 1, 2011,
http://www.fico.com/en/FIResourcesLibrary/NewYorkInsuranceForum2011-3_
Identify_Insurance_Claims_Fraud.pdf
.

“Fraud Detection Acid Test,” SAS Institute, October 4, 2012,
http://www.sas.com/knowledge-exchange/
risk/fraud-financial-crimes/fraud-detection-acid-test/index.html
.
18
“Operationalizing a Fraud Detection Solution: Buy or Build?”
Insurance & Technology
, August 20,
2012,
http://www.insurancetech.com/business-intelligence/operationalizing-a-fraud-detection-solut/
240005814
.
19
“Data Storage: Managing Unstructured Data in the Cloud: 12 Factors to Consider,” July 27, 2011,
eWeek
,
http://www.eweek.com/c/a/Data-Storage/Managing-Unstructured-Data-in-the-Cloud-12-Factors-to-
Consider-215018/
.
20
Digital data, the majority of which is unstructured data, is expected to grow from 130 exabytes to
40,000 exabytes between 2005 and 2020, according to a 2012 survey by IDC and EMC.
About Cognizant
Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process

outsourcing 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 in
-
dustry and business process expertise, and a global, collaborative workforce that embodies the future of work. With
over 50 delivery centers worldwide and approximately 150,400 employees as of September 30, 2012, 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
for more information.
subject to change without notice. All other trademarks mentioned herein are the property of their respective owners.

© Copyright 2012, Cognizant.
means, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is
1 Kingdom Street
Chennai, 600 096 India
Paddington Central
London W2 6BD
Teaneck, NJ 07666 USA
Toll Free: +1 888 937 3277
All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any
#5/535, Old Mahabalipuram Road
World Headquarters
Phone: +44 (0) 207 297 7600
Email: infouk@cognizant.com
Email: inquiryindia@cognizant.com
Fax: +91 (0) 44 4209 6060
Phone: +1 201 801 0233
India Operations Headquarters
Phone: +91 (0) 44 4209 6000
Fax: +1 201 801 0243
European Headquarters
500 Frank W. Burr Blvd.
Fax: +44 (0) 207 121 0102
Email: inquiry@cognizant.com
Okkiyam Pettai, Thoraipakkam
Credits
Author
Vinaya Kumar Mylavarapu, Senior Research Associate, Cognizant Research Center
Subject Matter Expert
Nipun Kapur, Director and Head of Analytics COE, Cognizant Analytics
Design
Harleen Bhatia, Creative Director
Suresh Sambandhan, Designer