Why New Approaches to Counter Claims Fraud May Not Work Jason McGrath, President, KLJ Computer Solutions | Mar 22, 2018

./iStock-488874360 (1).jpg

Using big data analytics to better understand markets, opportunities and risk is fast becoming a strategic differentiator in the insurance industry. Soon data-driven AI will effect dramatic changes in operations – automating key aspects of underwriting, billing, claims adjudication and settlement. But the potential of advanced analytics and automation to enable a strong competitive position will be only be possible when accurate and instantly accessible data are available across departments, partners and compliance organizations.

Focusing on fraud

Fraud is costing insurers big money. As much as 10 percent of property and casualty loss expense is due to fraud. For the five years ending 2015, that translates to about $34 billion annually.1 That’s a big hit to your combined ratio.

Fraud could come from virtually anywhere in the transaction chain – applicants, policyholders, or third-party services. It can run the gamut from misrepresenting facts on the application to claims for damages that never occurred. And fraud is on the rise. State fraud bureaus across the country are reporting increases in fraud referrals, cases opened, and convictions.

Traditional fraud detection doesn’t work anymore

As fraud schemes become more sophisticated, traditional tactics to combat fraud are no longer sufficient. Here are just three examples of why the old approaches need updating:

  • When fraudsters discern that a single rule drives a fraud detection process, they find a way to circumvent it. For example, once fraudsters understand that claims filed within 30 days of a policy’s inception set off a red flag, they find a workaround.2
  • Traditional approaches focus on detection after payment is made. This old “pay and chase” method is much costlier than uncovering fraud earlier in the claims lifecycle.
  • In the past, organized fraud rings would obtain policies from agents who had a reputation for skipping key underwriting questions. Today, with direct-to-consumer web portals, fraudsters simply find loopholes in the online application.

How to fight fraud in the age of insuretech

New technologies are a boon to insurers, but they also give rise to new fraud schemes that are harder to detect with a single approach. That’s why experts recommend using multiple techniques to fight fraud. When used together, they make it harder for fraudsters to counter, and they flag a higher volume and quality of claims. And unlike the old “pay and chase” approach, they flag claims much earlier in the cycle, saving cost up front.

These are the four key tactics that experts recommend using in concert3:

  • Anomaly detection: Create peer groups and look for actors that fall outside the norm.
  • Predictive modeling: Use past fraud data to detect similar patterns. Some insurers have seen a 20 to 50 percent reduction in fraud loss with this technique. But as mentioned earlier, fraudsters will adapt to this approach, so it’s ineffective on its own. 
  • Text mining: Use keywords – extracted from past claims referrals – to uncover fraud in unstructured data such as adjuster’s notes or first notices of loss. 
  • Social network analytics: Monitor social network posts for behaviors that a claimant would be unable to perform if the claim was true.

… But here’s the catch

Hybrid analytics is a fine solution to fraud detection, but if your data isn’t integrated it’ll be a heavy lift. While analytics platforms are evolving rapidly, access to source data is problematic.

Typically, policy underwriting, billing and claims are silos with weak connectivity between them. Even at the most basic level – establishing a single point of data entry, for instance – exchange of data between systems is often primitive and sometimes non-existent. Data disparity between systems can have a profound impact on efficiency and the ability to address important issues like fraud detection and accelerated settlement of claims.

Integrating with outside data sources is important, too. Insurers are looking beyond internal data to sense and respond to clues that spell fraud. Around 75% of insurers surveyed use both external data and analytics in their fraud mitigation programs.4

Seamless integration between policy, billing, claims, compliance systems, and third-party data ensures immediacy, accuracy, cost savings. So, how well connected are you?

1 Background on Insurance Fraud. iii.org. November 6, 2017.
2 Kuster, Kim. Combating Insurance Claims Fraud with Hybrid Analytics. Propertycasualty360.com. September 8, 2017.
3 Ibid.
4 Fannin, Todd. The Changing Face of Fraud. Propertycasualty360.com. September 20, 2016.

Jason McGrath is the President of KLJ Computer Solutions, Inc, starting his career in specialized custom software development in 1994. He has been involved in the insurance industry for over 20 years, assisting the technological needs of carriers, MGAs, and solution providers alike.