Pinpointing Where Predictive Analytics Makes the Most Impact for Insurers
By Stuart Rose, Market Strategy Director for Guidewire Predictive Analytics

Predictive analytics has proven its mettle over the past few years, and has become much more than a buzzword. In fact, businesses across markets and verticals can claim benefits from using it. Property and Casualty (P&C) insurance is no exception: A recent Novarica study, “Five Potential Changes in 2017 for Insurance,” indicates that a majority of insurers polled see predictive analytics as transformative to their business and a real driver of change in the industry. As a result, many are replacing their legacy infrastructure with new core systems that can support modern predictive analytics models to improve the quality of their decisions and protect profitability.

While most of us immediately think of predictive analytics as an invention born of the Digital Age, one could argue that predictive analytics has been around for quite a while — albeit in a more rudimentary form. For the past 200 years or so, insurers have used historical data to forecast what is likely to happen in the future. But the value of modern predictive analytics has increased, thanks to two main reasons: First, today’s analytical tools are decidedly more sophisticated; second, there is much more digital data to mine – and that’s when modern predictive analytics models really shine. Rather than yesteryear’s approach of basing forecasts on only a handful of variables, today’s tools synthesize and analyze big data to mine hundreds if not  thousands of predictive variables and influences. The result is a much more granular analysis of data.

Several areas in an insurer’s business stand to gain a lot from today’s predictive analytics tools — certainly underwriting and pricing, which have been the traditional areas of use for predictive analytics. These days, insurers base their pricing and determine risk according to a wealth of data that incorporates information from geo-spatial and telematics as well as traditional claims history. The amount and type of data collected ensures that solid pricing models are built, so that insurers neither price too high and lose customers, nor price too low and pay out more in losses than what they receive in premiums.

Marketing and claims are now emerging as potential areas of application as well. In marketing, predictive analytics can identify up-sell (such as additional coverage on an existing policy) or cross-sell (such as bundling auto and home insurance) opportunities, or whether to target certain individuals in specific regions for new offers. Did the policyholder change his address for his car insurance, and could that indicate an opportunity to sell home insurance? Do public records show that a prospect or policyholder recently bought a house, and is that house in a flood plain? Armed with information gathered from internal and external data, the insurer can instantly determine whether or not eachpolicyholder or prospect would be a risky or profitable investment.

Predictive analytics is also proving its mettle in claims — perhaps the biggest area of expense for insurers. Here, insurers use predictive analytics to analyze structured and unstructured data throughout the claims cycle, from FNOL to payout. When a claim comes in, predictive analytics can help determine the claim size, and whether it comes from a loyal customer on a broken windscreen, for example. With that information, an insurer might simply pay out without adjucating it, thereby reducing the insurer’s expenses and improving customer satisfaction.

Predictive analytics can even quickly determine whether the claim might involve some litigation — and, if identified soon enough, this could mean the difference between a senior adjuster negotiating settlement for a claim early on or wading into a lengthy and expensive court battle.  

Perhaps the most compelling use case for predictive analytics in claims is in fraud detection, a massive problem in insurance. For example, a policyholder might claim hail damage to his roof, but historical weather data collected and analyzed using predictive analytics models may say otherwise, pointing to a clearly fraudulent claim. Certainly, a small number of claims are fraudulent as compared to the number of legitimate claims filed, but nonetheless, fraud continues to be an expensive problem for insurers. While one-off fraudulent claims seem manageable, organized claims fraud is gaining steam and becoming big business. Insurers are battling this with a combination of tools, including predictive analytics models, to proactively predict fraud before larger losses occur. 

Data is the lifeblood of P&C insurance.  And there isn’t an insurer out there that doesn’t wish it had better data.  Given the amount of data insurers already have, how can they get the most value of what is hidden within it?  Well, now the ‘machine’ can do the digging and find the insight for insurers, resulting in reduced losses, minimized inefficiencies, and a path to greater profitability.

Stuart Rose is market strategy director for Guidewire Predictive Analytics, responsible for thought leadership and marketing content for applying big data and analytics within the insurance industry.