Sometimes innovation comes from seeing how somebody does something in another business and then applying the same idea to what you do.
That’s how William Dibble, senior vice president of Infinity Property & Casualty Co., came up with the idea of “scoring” insurance claims from customers to look for signs of fraud. His model: the way lenders score credit applications. Infinity uses predictive analytics technology provided by IBM to spot potentially fraudulent claims and speed the payment of legitimate ones. Infinity’s experience illustrates the value of using data analytics to transform the way a company does business.
Editor’s note: Today, at an event in New York called the Smarter Analytics Leadership Summit, IBM is talking with 100 business leaders about the potential for harvesting insights from Big Data to create value for their businesses and for society. In addition, we want to open the conversation to many more people via this blog and Twitter, at #IBManalytics and #Big Data. For the video simulcast, click here.
Infinity, based in Birmingham, Alabama, specializes in providing auto insurance for drivers who present higher than normal risks and pay high rates for comparable coverage. Because of the business the company is in, spotting fraud is even more important than it is to other insurance companies.
Dibble started off by using predictive analytics to analyze statistical information stored in traditional data bases, but now he’s branching out into the Big Data realm. The company is starting to use text analytics to spot evidence of potential fraud in written reports. Infinity is also applying analytics to other parts of the business, including direct marketing. Says Dibble: “There’s no end to what data analytics is capable of doing. Every time we come up with ideas, holy mackerel, it just leads to others!”
The claims fraud system increased the success rate in pursing fraudulent claims from 50% to 88% and reduced by as much at 95% the time required to refer questionable claims for investigation. Dibble also points out that, partly due to the system, Infinity has made an underwriting profit every year since the system was installed–a rare feat among auto insurers in this economic climate.
Here’s how the text analytics technology works: Claims adjusters write long reports when they investigate claims, telling the story of the incident from beginning to end. Some of these reports can amount to thousands of words. Hidden in the text are clues that the adjuster may not notice or facts that he or she might forget. But the computing system, based on rules, can spot evidence of possible fraud. For example, if someone claims that their car caught fire but the narrative indicates that they took everything of value out–such as golf clubs–that might indicate they torched it. Also, people who commit fraud tend to change their story over time. The system can spot that kind of discrepancy.
Infinity has just begun using text analytics on a limited basis, but Dibble hopes to have it rolled out in a big way this year. “This is really big data being put to work,” he says.