Decreasing the Cost of Fraud

Although it is estimated that up to 10 percent of all claims are fraudulent, insurance companies are able to uncover only 0.5 to 1 percent of the suspected fraudulent cases. The potential for improvement is huge. In the past, insurance companies were not interested in detecting fraud; they simply would accept fraud, increasing premiums to make up for the increased costs of claims.

Two trends are changing this now. Increased pressure from government and a growing disapproval among the public are fostering a widespread belief that fraud is a crime and that it is unacceptable that other citizens should pay for these costs. In addition, reducing the cost of fraud allows insurers to offer more competitive premiums.

Technology also is coming to the rescue. Predictive software can help insurance companies uncover two to three times as many instances of fraud, whilst enabling them to staff the first line of fraud detection with the people who do it best: the branch and call-center teams who speak to customers every day. Such a strategy, if implemented correctly, can enhance fraud detection and customer relations, and does not require vast investments in infrastructure or personnel.

The key lies in a program that can be integrated into existing claim-handling systems and provides claim handlers with advice alerts when they enter new claims. This advice is based on experts' insights into fraud, combined with advanced analytic software. Based on this, the claim handler can decide to settle the claim immediately, thereby offering improved service and reducing costs, or to perform a normal investigation or send the claim to the fraud detection department.

Most fraud cases rely on that most unquantifiable of human instincts, the gut feeling. Many fraud experts say that they can spot fraudulent claims instantly; the time spent on a case is to substantiate their instincts. These instincts, of course, are based on considerable experience, but they can be boiled down to a handful of key risk indicators that can show how each case should be handled.

Although this knowledge is available within the fraud department, other areas of the company do not draw on it widely. Institutions can gain considerable value by interviewing experts and building their knowledge into the claim-handling process, to flag those claims that require further investigation.

This process, which I call mind mining, can help determine the handling of a great deal of claims, but it is not the whole picture. Many fraud profiles are unknown to the experts but can be revealed by analyzing historic fraud data to predict future trends, a technique known as predictive analysis. Entirely new classes of fraud can be uncovered using predictive analytic technology, increasing annual fraud detection rates two- or three-fold.

The fruits of the two processes can be developed into a range of fraud profiles or sets of criteria that establish whether new claims might be fraudulent. If a claim were to match any of these profiles, it could be sent to the fraud department for further investigation.

The next stage is to put into place a process that allows fraud profiles to be updated on an ongoing basis, enabling the understanding of new fraud trends as they develop. Predictive analytics also is effective here. Fraud experts can delve into claim databases and use the technology to spot new trends. This is a simple, ongoing feedback loop of analysis and profiling by fraud experts, followed by deployment and action within the call center.

Customer Service

Insurers regard the speed with which they can respond to customers' claims as a visible way to distinguish themselves from their competitors. Setting up a fast-track claim process provides obvious customer service and sales benefits, while allowing companies to reduce claim-handling costs, by up to 40 percent in some cases. The success of this strategy relies on an early understanding of which claims are unlikely to be fraudulent. Without this understanding, insurers leave themselves exposed to fraud.

Many claim handlers and customer service representatives lack the expertise to recognise fraud. Companies must consider carefully how CSRs can match claims to fraud profiles. In busy call centers or branches, it would be highly impractical, and unnecessary, to present CSRs with lengthy fraud questionnaires each time they register new claims. Analytic and scoring software can reduce the claim-handling process to minutes with a handful of leading questions, and provide immediate recommendations on whether a given claim is potentially fraudulent and should be handled by the normal claim-handling process, or whether it can go through the fast track.

Imagine that a motor insurer's claim-handling system has been programmed to flag any claims involving claimants under 25 years of age, living in Chicago, and driving red or sports vehicles. John, a 46-year-old from East St. Louis, calls in to make a claim for damages to his car. The first half of the profile can be discarded altogether, but John confirms that he drives a red MR2. While John is still on the phone, the system advises the CSR that, on the basis of the car type and color, the claim might be suspicious and requires further investigation. The CSR can advise the customer that an investigator will be in touch soon, and end the call.

If however, John's car is blue, the system may determine that the claim is likely to be watertight and can be settled immediately. Not only can this reduce the burden on the fraud department, the ability to offer genuine claimants settlements on the spot opens the door to a whole range of new sales possibilities: an insurer is almost guaranteed a hit with a customer whose claim has just been approved.

One-stop claim handling need not be as ambitious and demanding as its name might suggest, and it broadens customer service, sales, and marketing possibilities. Given careful investment in the right technology, and an emphasis on implementing the closed-loop processes outlined above, insurers can catch up with fraud perpetrators and provide better service to customers. How's this for a vision of insurance in the 21st century?

Marcel Holsheimer is vice president of product marketing at SPSS, a provider of predictive analytic technology.