(Editor's Note: This article has been contributed by Stuart Rose, global insurance marketing manager at SAS.)
Insurance companies have implemented new claims management systems to improve the claims process, yet loss ratios continue to rise and fraud still occurs. To remain competitive, insurers need to consider applying predictive analytics across the claims life cycle. By looking at claims data in its entirety, insurers will be better positioned to optimize loss reserves, increase productivity, and root out fraud.
Fraud is just one example where analytics across the claims cycle is beneficial. Analytics can improve the bottom line by:
- Enhancing recovery efforts. By scoring claims at every stage of the process and employing text analytics, insurers can quickly find and monetize salvage, subrogation, and third-party opportunities.
- Accelerating settlements. From a customer service perspective, the pressure is on to settle quickly. Analytics helps insurers avoid overpaying, without slowing down settlement speed.
- Benchmarking claims with confidence. Estimating the size and duration of a claim when it first comes in is often a guessing game. Predictive analytics gives insurers the information needed to accurately predict loss reserves despite the often long-tail nature of many types of claims.
- Managing resources and litigation wisely. Robust analysis helps insurers automate the process of assigning the right adjusters to the right case and calculating litigation propensity scores.
Finding Suspicious Claims Before Payment
There could be a bit of a silver lining to the recent NICB report. It is possible the uptick in suspicious claims is because insurers are getting better at spotting them. Yet, it is one thing to spot suspicious claims, and quite another to identify them before the payment has been sent. In addition, traditional fraud solutions are too easy to game because they detect fraud with manual or automatic business rules. With an estimated 10 percent of all claims being fraudulent, insurers need a better approach. Fraud analytics uses traditional rules and anomaly detection along with advanced analytics and social-network analysis for a hybrid approach that is particularly successful at finding the common linkages that even the most sophisticated fraud rings can’t cover up.
Finding the Right Payee
Subrogation is one of the few processes that has not been optimized, in part because the process is so manual. You need to be able to automatically score a claim at every stage of the process to look for salvage, subrogation and third-party opportunities. Many recovery opportunities remain buried because the information is hidden in a claims narrative. Text analytics is a critical tool since it automates the process of combing through unstructured data to find phrases that typically indicate a subrogation claim. A classic example of this involves the troubles with the brakes on Toyota vehicles. One major insurer was able to analyze its database and notify the National Highway Transportation Safety Administration of a spike in claims. Finding issues quickly can both help manufacturers remedy problems and provide insurers with subrogation options.
Speeding Settlement Without Overpaying
A colleague of mine chuckles when she retells the story of the dramatic uptick in the installation of in-ground pools in Miami after a hurricane tore through the area. It turns out that some claims agents, eager to keep customer satisfaction high, cut checks too quickly. However, if insurers treat a claim suspiciously or otherwise drag out the process, they risk irreparable brand damage—often made worse by the lightning speed with which social media can make a company look cruel.
Predicting Size, Duration for Better Loss Reserving
You can’t always control how much money you will need to pay out in claims, but it certainly helps if you can estimate the costs, especially in relation to loss reserving. It can be particularly tricky for long-tail claims areas, such as liability and workers’ compensation. Consistent claims management from the beginning of the claim life cycle can reduce the need for incremental increases of loss reserve. Analytics can more accurately calculate the loss reserve by comparing a loss with similar claims. In addition, whenever the claims data is updated, analytics can reassess the loss reserve. Improved loss reserving accuracy allows insurers to move funds from bulk reserves into more flexible investments.
Here’s an example of how that works. A driver crashes into a tree. As soon as the insurer has the police report (unstructured data that is best suited to text mining), it takes that data and analyzes it against variables from similar crashes: the car’s speed at impact, age of driver, make and model of car, the number of passengers, etc. The insurer then uses this information to estimate the loss reserve.
Analytics can also be employed by claims managers to measure the effectiveness of the claims handling process—in particular, adjuster efficiency. Traditionally, adjuster productivity had been based on whether an adjuster closed more claims than were opened in a reporting period. That crude method can be replaced with a key performance indicator system that measures adjuster performance based on customer satisfaction, overridden claims settlements, and other related metrics.