Improving claims processing isn't a simple task. Thousands of claims, with hundreds of different access points, and millions of records create ripe exposures for massive fraud. The Coalition Against Insurance Fraud conservatively estimates that insurance fraud costs $80 billion a year in stolen claims. And that number is grossly higher when you start adding in the lost productivity of businesses and the cost to investigate and prosecute against fraud.
Like many other mid-size insurance businesses, and as a specialty insurance provider, we at Infinity Property & Casualty struggled with effective fraud management. In 2007, we sought to re-engineer the claims processing function by developing a smarter system, tagging suspicious activity from the onset of the claim reporting cycle. We realized that automobile insurance claims could be scored in the same way as consumer credit applications, and that technology could significantly enhance that process by making specific inferences on behavior. By using predictive analytics to govern our approach, we developed a process to assign different "scores" on fraud probabilities when claimants first report an accident. By using a rules engine that automatically scores claims based on its fraud probability, we could forward suspicious claims into the hands of investigators within a day or two for deeper analysis. As a result of using this technology, not only have we slashed the time it takes to identify potential fraudulent claims within 24 hours -- it used to take between 30-60 days -- and we have been more successful in catching fraud claims.
An appreciating investment
Evaluating claims has long been a judgment-heavy, labor-intensive part of the business. In the first month of using IBM's SPSS analytics-based system, our company has been able to increase subrogation recovery by $1 million; after six months, the savings rose to $12 million. Subrogation is the process of getting money back from the company or people responsible for the damages.
Using predictive analytics, we had the unique opportunity to close a hole in our pocket where money was leaking out steadily. With so many pressures being put on insurers -- including to maximizing profits, boost shareholder value, and operate in a more consolidated and competitive marketplace -- the ability to minimize losses is absolutely critical to survival. Not to mention, the application of a rules-based system ensures increased reliability and protects against human judgment and error.
The beauty of this investment is that it appreciates over time. For every dollar spent on the IBM SPSS analytics product, we are capturing $403 dollars worth of lost money. Moving forward, we want to continually update our models to further improve accuracy to pinpoint the underlying cases of fraud.
A commitment to better service
At the end of the day, not only do we want to protect our business, but keep our customers happy during their interactions with us. For example, by using analytics to identify claims least likely to require investigation, and handling those cases internally, we are now processing claims at a rate of 35.5 percent faster than before. We have empowered staff to take on more accountability for decisions that adjusters in the field would have ordinarily made, as well as provide "express treatment" to customers with legitimate claims and put them on a faster path to resolution. In the past, without having such as effective screening process in place, the payment process took considerably longer.
By better allocating our resources, we are putting the focus back on service, where it belongs. And customers have noticed. Customer satisfaction levels have increased year over year since 2007, with the more efficient handling of claims. In addition, we have grown our business at a clip of over 10 percent, producing profit in a challenging economy.
Building a smarter claims system takes considerable skill, resources and time. But the benefit of doing so is making the claim processes less cumbersome, boosting customer service and generating profits, opposed to bleeding losses. Here are a few suggestions to approach building a smarter system for fraud:
Think of the big picture: If you are looking to seriously address a fraud management problem, remember to think about how a new system can be applied across the enterprise. You'll have to corral the business to pay for the investment. But by taking a long range view, the results will be put into better context.
Start small and then expand: Once you understand how you can apply the power of predictive analytics, you can expand the scope of work. For example, here at Infinity we are exploring the use of an advanced text mining capability to analyze the content of police reports, medical reports and other documents related to accidents to refine our fraud management capabilities. Outside the area of fraud management, we are looking to use analytics intelligence to enhance processes associated with direct marketing.
Don't try to do it all: A new rules-based engine will not be a panacea for all your claim processing work, however, it can significantly make it more efficient. Like with most things in life, take small, measured steps to see the best results. Whether it is used for fraud reduction, customer service or cost control, technology can be a strategic differentiator for the business.
About the Author: William "Bill" Dibble is senior VP of claims operations, Infinity Property & Casualty He can be reached at (205) 803-8364.