Business Analytics Blog

Business Analytics Blog

Opinions expressed on this blog reflect the writer’s views and not the position of the Capgemini Group

Cops and Robbers in the Digital Era by Rob Fitton, Alexa Morrison and Nigel Lewis

Improving existing operations to reduce risk, increase margins and yield competitive advantage is more important in the current market than ever – and we believe that analytics can provide the answer. This week Capgemini’s Business Analytics team are releasing a daily blog on some of the key use cases of Operational Analytics – today’s topic is Fraud Analytics.

Cops and Robbers in the Digital Era

The game of cops and robbers has been around for centuries. Wherever there is a corporation or government with money to extract, there will be criminals trying to extract that money.

In the Victorian era, criminals would climb down a rope ladder hoping to crack open a safe with a skeleton key. In the digital era, things are a little more advanced; instead of cracking safes with skeleton keys they are cracking passwords with encryption keys. Just as the criminal has adapted to the wonders and perils of the digital era, modern security and defence systems are close on their trail. One of the most effective weapons in this battle is the cutting edge discipline of fraud analytics.

Fraud analytics is not just used for tackling digital crimes; it reaches far beyond into the physical world. It is used by banks and insurance companies to tackle fraud and by governments to tackle terrorism. With cyber crime globally costing organisations around $400 billion annually, defences such as fraud analytics have never been in higher demand.

So how are organisations and nations exploiting the power of fraud analytics to defend and protect themselves against financial losses? One technique that is widely employed is known as social network analysis (SNA). This system is widely used by tax authorities, insurance and banking institutions, as well as military intelligence agencies.

One well known implementation of SNA, was initially developed by a private sector defence company to identify high risk targets such as terrorist cells by extracting datasets from multiple sources. It then connects these datasets together to provide a unified view of a suspect’s life to uncover links with other terrorists, criminal safe houses and illegal bank accounts. This joined up “connected” view of a suspect is also known as the suspect’s social network. Networks such as these have even been used to demonstrate links between the suspects involved in the 9/11 attacks carried out in New York. This powerful analytics technique has since been adopted in other sectors such as insurance and finance to detect and identify potential fraudsters. 

For example, when a customer takes out an insurance policy, their social network can be compared with the social networks of other customers with similar characteristics. If they are an IT consultant, aged 30, from the midlands and earning £20,000 a year, they can be compared with other people of the same age, profession and salary. A picture is built up of the asset values of the typical person in this profile and whether the new claimant fits into it. If the norm for customers fitting this profile with similar lifestyle characteristics is to take out a policy worth on average £100,000 but this claimant is attempting to take out a policy valued at a million, the chances are this individual would be set aside for more thorough examination.  



SNA can also be used in combination with other analytical techniques such as data mining, cyber and machine learning. One such use is to tackle a pattern of fraudulent behaviour known as identity theft. Risk networks are used to detect the likelihood that an online applicant is attempting to obtain credit with someone else’s details. An assessment is carried out on the information provided by the applicant during the present with information linked to the same applicant in the past. For example, machine learning systems can compare thousands of connections in the individual’s social network such as linked family members, legitimate bank accounts, phones, emails, businesses and addresses. Discrepancies between past and present networks are highlighted. Then algorithms are utilised to model how likely these discrepancies are to be genuine changes of circumstances or whether they resemble patterns associated with identity theft. The applicant can then be accepted or rejected accordingly.

Using this analytical method called “social network analysis” enables companies, organisations and law enforcement to catch fraudsters, criminals and even terrorists. Fraud analytics comprises a huge array of exotic methods and technologies in addition to social network analytics such as Bayesian probability, clustering, random forests and capture/recapture.

So the next time you think of the game cops and robbers, think beyond the wooden truncheons and handcuffs and delve into the digital world of fraud analytics.



If you would like to know more about Fraud Analytics, our Operational Analytics offer or where Capgemini has delivered this capability before – please feel free to contact one of the authors.

About the author

Nigel Lewis
Nigel Lewis
Nigel leads the Capgemini Consulting’s 35 strong Business Analytics team, which delivers analytical, operational and strategic modelling solutions to clients. He has 18 years consultancy experience as well as 8 years experience in the UK gas industry. Nigel has successfully managed complex projects in both the public and private sector, including capacity modelling, simulating supply chain operations, strategic business modelling to support future policy decisions, and implementing complex demand forecasting systems. Nigel is currently focussing on the development of Capgemini’s customer analytics and analytics advisory services.

Leave a comment

Your email address will not be published. Required fields are marked *.