Insurers Starting to Live Up to Data-Driven Reputation

The insurance industry has a reputation for being data-driven. There’s no emotion involved, just cold, hard data. But in reality, there’s more than a little gut decision making happening at most insurance companies. Sure, insurers use tools like actuary tables to determine life expectancy, but even these are based on aggregate data and provide only generalized estimates. As a result, safe-bet customers are subsidizing riskier ones, and insurance companies are leaving money on the table.

To put the scope of the problem in context, consider that insurers paid out over $27 billion for natural disaster claims alone in 2015. Add up claims paid across all types of insurance — health, property, auto, etc. — and the number is in the hundreds of billions of dollars annually. If even just a fraction of a percentage of claims paid are associated with inaccurate risk assessment, that’s still billions of dollars being wasted every year.

But the insurance industry is starting to live up to its data-driven reputation. Insurers are increasingly investing in emerging technologies, many based on open source software, to more effectively collect, process and analyze data from both traditional and new data sources in volumes never before possible. This includes taking advantage of data created thanks IoT technology, such as connected car data and fitness tracker data. Both developments allow more accurate decision-making based on both historical context and the particulars of any given case rather than relying on generalized risk assessments of populations of similar individuals and organizations.

While still early days, data science is poised to significantly change the ways insurers compete for new customers and maintain profitability in a complex world, according to McKinsey & Company. “In the future, the creative sourcing of data and the distinctiveness of analytics methods will be much greater sources of competitive advantage in insurance,” write McKinsey’s Richard Clarke and Ari Libarikian in a recent article. “New sources of external data, new tools for underwriting risk, and behavior-influencing data monitoring are the key developments that are shaping up as game changers.”

The case of Marsh ClearSight is illustrative of the transformation taking place in the insurance industry. The Chicago-based company offers a cloud-based risk management and claims administration platform that, among other things, helps insurers better predict likely claim outcomes and recommends preemptive actions to more effectively deal with them. Marsh ClearSight CTO Mark Pluta says many of the company’s clients — insurers and claims departments — are being asked to adjudicate more and more claims every year, but haven’t enjoyed a corresponding increase in budget or resources to do so.

Many are turning to data science to keep pace with the demanding workload, he said. In one example, Marsh ClearSight is developing analytical capabilities that enable insurers to quickly and accurately predict likely settlement outcomes of new claims. With that insight, insurers can put their best claims adjusters on those cases that are likely to end up in larger payments and more quickly adjudicate the less costly claims. By more intelligently allocating their top resources — top claims adjusters — insurers can not just better keep up with the growing workload, but actually improve outcomes in the process.

Such a service wouldn’t be possible without a number of important capabilities. First is the ability to store and integrate huge volumes of data from multiple sources. In the case of Marsh ClearSight, this includes historical claims data, financial transaction data and even unstructured text such as notes from claims adjusters in the field and call center interactions. Second, it requires moving beyond basic business intelligence and reporting to more sophisticated data science and predictive modeling techniques. Third, the system must deliver insights about likely claims outcomes in ways adjusters can consume and take action on them. And finally, the results of these actions must be fed back into the system to continually improve the predictive models.

For Marsh ClearSight, this required investing in both new technology and new processes. From a technology perspective, Pluta and team adopted Pivotal Greenplum, a massively parallel processing analytical database, to handle the analytics workload. From a process perspective, Pluta said the team now takes an agile, iterative and collaborative approach to developing predictive models, a job that is never really “finished.” There is always room for improving results and uncovering new insights, he said.

Marsh ClearSight is currently piloting the new analytic capabilities with a select set of clients to measure, refine and improve the predictability of outcomes with a goal to open up the service to all clients in an upcoming release.

Of course, there are many types of insurance — home, life, auto, health, corporate, etc. — so the application of Big Data and data science will take different forms depending on use case. Marsh ClearSight is just one of potentially hundreds or even thousands of examples. But the larger point is clear: Insurers that both adopt Big Data technologies and data science practices and intelligently apply them to high-value business challenges and opportunities will be the big winners in this transforming industry.