The application runs off a predefined, yet highly customizable, predictive model that is honed from insurance best practices gleaned from field experiences. Speeding up claims processing and combating insurance fraud are foremost in the minds of insurers today, according to Marcel Holshiemer, vice president of vertical markets at Chicago-based SPSS.

Insurance firms are concerned with three things: improving customer-facing processes, cutting costs and reducing fraud, he said.

SPSS’s software helps by streamlining the painfully long claims handling process with an eye to fast tracking low-risk claims. According to Holshiemer, the goal of insurance companies is to fast-track 40-60% of claims in order to increase customer satisfaction.

Insurers want to compete not on price, but customer satisfaction, he said. Insurance companies that can decide whether a claim will be paid within 15 minutes have a definite competitive edge.

A natural by-product of fast-tracking is a reduction in claims handling costs: of up to 20-40% according to Holshiemer. He cites industry statistics that show insurance companies spend 13% of their turnover on the manual graft of claims processing.

Automating and streamlining the process cuts out on huge volumes of calls to the call center. Around two-thirds of calls are just about claims status, which is a huge load and cost.

Holshiemer added: Statistically its been shown that the longer its takes to settle a claim, the higher the average amount is paid out. Hence it’s in the financial interests of insurers to settle as quickly as possible.

The most valuable benefit of using predictive analytics in claims processing is fraud detection. Research estimates that around 10-15% of claims are fraudulent in nature.

The software acts like an early-warning system, providing a real-time assessment of a current claims to determine how they should be handled next; for example, if it should be fast-tracked, handled normally or sent to an investigation team as suspected fraud case, Holshiemer said. Because the fraudsters are quite creative, we’ve made it easy to bend the rules in the system to keep on top of newer forms of fraud.

Holshiemer said SPSS holds an edge over other fraud detection solutions in the market because of its competencies in text analysis and research. Around 80% of claims information is held in textual form, things like accident descriptions. Our text analysis software helps to dig out patterns that detect fraud, he said.

SPSS’ experience in designing and creating surveys also helps by automatically pushing out smart questions at customer touch-points (like call centers) to seek out inconsistencies in the claim that point to fraud. Holshiemer said that SPSS’ survey software is already used by almost all the leading market survey firms today.

PredictiveClaims is the third application built top of a foundation of SPSS’ core analytics, data mining and statistical technologies.

The company has already released other Predictive-branded analytic solutions for marketing and call centers that are primarily aimed at the financial services, telco and insurance sectors.

These [two applications] were all to do with CRM – things like cross-selling and customer retention. Now we’re extending out to risk-related applications with PredictiveClaims, Holshiemer said.

Of course there are real business benefits of linking these three Predictive applications on a single platform. For example, if you’re able to predict claims fraud you can avoid cross-selling to these customers before not after the fact.

SPSS says that eight of the top 10 global property and casualty insurers are already customers of its analytic software. The company has several early adopters of PredictiveClaims already lined up in Europe.

Holsheimer said that while the initial release of the application aimed squarely at automotive and home claims environments, SPPS plans to roll out support for other commercial insurance segments later this year.

SPSS would not release formal pricing saying the cost of implementation depends heavily on the size and complexity of implementation.