Improving fraud detection rates through predictive analytics

Throughout the UK, insurance fraud is increasing. Research by Synectics Solutions published in the National SIRA fraud trends report this year showed that claims fraud in the UK increased by over 32% and policy fraud by just over 18% for the first half of 2017, compared to the same period in the previous year.

Moreover, the increasing level of identity and impersonation fraud has also increased exponentially, rising from 9 cases during the first half of 2016 to 6,059 cases for the same period of 2017, amongst members of the National SIRA syndicate. In fact a recent UK Government report from the Insurance Fraud Task Force published in 2016 put the cost of fraud from insurance at somewhere around £3 billion.

Furthermore the problem is far wider than the UK and is generally considered to be a global issue.

An Insurance Europe Report published in 2014 documented many of the fraud issues that insurers around Europe are facing that are costing them billions of Euros.

THE CHALLENGES INSURERS ARE FACING

The range of tactics used by those that are looking to perpetrate fraud to falsify claims and policies to reap the rewards of unauthorised funds is quite staggering. Whether dealing with false claims and identities, staged incidents, withholding of information, organised fraud rings, or ghost broking, insurers are being confronted with an unprecedented level of crime. This needs to be identified to prevent losses from this kind of activity destabilising their business and reducing their, already, slim profit margins.

Historically insurance companies have deployed a variety of data matching and rule based decision making solutions to help them identify, detect and address this problem. These solutions have been tremendously effective at arming these companies with the tools to defend themselves against this onslaught. And today the mainstay of many major underwriting and claims management fraud defences use rule based decision making very effectively to help them reduce their losses to fraud.

"...the increasing level of identity and impersonation fraud has also increased exponentially, rising from 9 cases during the first half of 2016 to 6,059 cases for the same period of 2017"

However, the volume of transactions that insurers are now having to deal with has created a number of factors that have contributed to those in this sector feeling that more needs to be done to help stay on top of this issue.

Overloaded investigation teams, high false positive rates, inability to effectively prioritise investigations - along with the need to speed up decision cycles to remain competitive, have all meant that insurers are now turning to predictive analysis models to help them augment and improve their fraud defences.

With limited resources for fraud investigation teams, hiring more staff to conduct manual investigations is an expensive and inefficient option outside the reach of most organisations.

IDENTIFYING FRAUD QUICKER AND EARLIER IN THE APPLICATION CYCLE

To help our clients keep pace with this and respond to these challenges, Synectics recognised that there was a need to enhance the ability of clients to improve their fraud defences, help them to improve their detection rates with less resource, identify fraud cases earlier, and assist investigations teams with better prioritisation of referred cases.

“Through its ability to process huge volumes of data and apply a blended array of algorithms and modelling techniques, Precision now allows insurers the ability to make risk decisions quicker.”

The rich data sets that insurers have built up over time coupled with the ability to deploy appropriate analytical models meant that predictive analysis offered an opportunity to identify and prevent fraud much more effectively than ever before.

Synectics Data Scientists realised that clients’ rich data, linked to the intelligence available within the National SIRA database - a syndicated resource contributed to by over 100 financial services providers, insurers, retail providers and telecommunications companies, provided the perfect opportunity to deploy a range of predictive models to help identify more fraud, categorise risk better and deliver fraud propensity scores to specialist investigators in real time.

“As Precision was developed as Software-as-a-Service (SaaS) it makes it much more cost effective and removes the need for organisations to recruit in-house data science expertise or deploy expensive technology. ”

So, Synectics invested in the development of its Precision solution to provide an enhanced suite of predictive analysis services, capable of delivering all the predictive insight clients need to establish a more proactive approach to fraud prevention.

Through its ability to process huge volumes of data and apply a blended array of algorithms and modelling techniques, Precision now allows insurers (and other financial services providers) the ability to make risk decisions quicker and more accurately than ever before, in addition to improving their workflow and fraud detection rates.

By calibrating Precision by training the various data models within the system’s software, potential fraudulent business can be identified much earlier in the customer cycle and prioritised, based on risk. This allows investigators to work the highest priority fraud cases first, thus improving detection rates and increasing efficiencies and investigation capacity.

Additionally another concern for organisations is how they continue to develop technology without the need for large investments, in terms of budgets for hardware, data science expertise or system maintenance. As Precision was developed as Software-as-a-Service (SaaS) it makes it much more cost effective and removes the need for organisations to recruit in-house data science expertise or deploy expensive technology.

CASE STUDIES: A MULTI-LAYERED APPROACH TO FRAUD DETECTION

The following case studies consider some of the results that were achieved during the pilot phase of the Precision service as well as a live use case from major insurance provider esure .

"Predictive analytics in the prevention of financial crime will become increasingly essential in proactively identifying and combating risk.”

CASE STUDY: IMPROVING THE ABILITY TO DETECT COMPLEX FRAUD CASES

From its earliest moment Precision has proved its effectiveness in being able to improve clients’ ability to detect some of the most difficult to identify fraud typologies, such as first party fraud, in comparison to traditional fraud detection methods.

Synectics found that be leveraging intelligence from within the National SIRA database and combining this with Precision’s powerful predictive analytics tools, they were able to increase their client’s ability to identify fraud by as much as 300%, when compared to more traditional fraud identification methodology.

Working with a major finance client Synectics were able to calibrate Precision in such a way that it was designed to align with the company’s specific appetite for risk. The results of this deployment resulted in the company being able to reduce the amount of investigations they were having to undertake by as much as 50% and increase their ability to identify fraud by 30% in comparison to their usual operational model.

Synectics were able to increase their client's ability to identify fraud by as much as 300%

CASE STUDY: PRECISION REVOLUTIONISES FRAUD INVESTIGATIONS FOR LEADING UK INSURER ESURE

Precision allows esure to enhance their existing fraud prevention and detection capability. The combination of manual referrals, cross matching clear and suspect claims and policies and other intelligence, integrated with the system-driven profiling alerts from Precision, allows their investigation teams to escalate fraud concerns to a special investigations team quicker and with more confidence. The seamless integration of the data analytics within their existing SIRA system has ensured a fully integrated workflow with rules-based claim fraud detection and intelligence-based data matching.

1 in 6 of esure’s claims, involving suspected crash for cash scenarios retained for investigation within their Special Investigation Unit is now sourced from Precision.

75% of cases accepted by esure’s Special Investigation Unit are characterised by high (or very high) Precision scores, even where the solution is not the primary investigation source, evidencing the overall accuracy and value of the analytics in supporting referrals generated from other detection layers. esure has also used Precision to recalibrate their decision model for identifying false positives with the aim of reducing referrals. The initial results of the recalibration are extremely positive.

Precision Business Development Manager Osman Khurshid said:

“Through its ability to process huge volumes of data, and apply a blended array of algorithms and modelling techniques, Precision enables a shift in fraud strategy from one of detection to one of prevention. Precision takes Synectics’ existing customer data and applies sophisticated analytical techniques and machine learning algorithms to generate previously unseen insights into client data.

“Because the solution was developed as a service model, Synectics can now offer companies a truly world class - and affordable - data science service with our cutting-edge software and service.

“Predictive analytics in the prevention of financial crime will become increasingly essential in proactively identifying and combatting risk. It provides access to leading edge data scientists, combined with machine-learning functionality, which offers a sophisticated multi-layered approach.”

Matt Gilham, Head of Financial Crime at esure said:

"While we already employ a range of procedures and technologies to identify claims fraud, Precision allows us to quickly implement enhanced, technical machine-learning capability to drive incremental benefits and improve operational efficiencies through seamless integration with our existing workflows.”

FURTHER INFORMATION

If you are interested in finding out more about how Precision can be used in your organisation to help prevent, detect and investigate fraud, then please call 01782 664000 or email info@synectics-solutions.com