Weather and AI make for a perfect storm in the insurance sector

How weather data can help streamline the claims process

The weather has always challenged our best and brightest—from Aristotle and his early theories of Meteorologica to the legions of engineers and scientists at organizations such as IBM’s The Weather Company.

The job isn’t getting any easier. Weather will always be a highly complex scientific phenomenon, but climate change now poses an array of increasing risks that scientists and actuaries alike are racing to comprehend. No industry (apart from the oldest industry, agriculture) wrestles more with these issues perhaps than the $5 trillion global insurance sector.

For instance, 2017 was the costliest year ever recorded for weather disasters, which racked up $353 billion in economic damages. It was also the third warmest year in recorded history. These are trends—and risks— that will likely grow in scale and complexity in the coming years.

But the technologists have some powerful new allies on their side: machine learning, predictive analytics, and artificial intelligence. Investment in insurance-technology startups firms more than tripled last year to $2.6 billion, with some of the most powerful applications in weather. Here are just a few that show how technology can convert risk into opportunity:

Storm tracking and alerts

Low-cost rooftop IoT-enabled sensors are now allowing insurers to track hailstorms house by house, a vast improvement over radar, which can only locate hail damage within several miles.

One promising example: Boston-based Understory’s rooftop device, shaped like a steel salad bowl, is the workhorse device in a rooftop network installed in six U.S. cities, and now working for several major insurance brands. One of those companies estimates that the technology has already saved it as much as 20 percent in claims by improving claim accuracy after recent storms.

After a storm, the presence of accurate hail data on your block can eliminate ambiguity as to who gets paid out and how much The same technology can be used to alert residents of an impending hailstorms—in come cases avoiding damage claims altogether.

Automated claims

Assessing and resolving claims has always been an unwieldly discipline in the insurance sector. How many adjusters and surveyors are needed in the aftermath of a big disaster to verify claim accuracy?

Nascent AI applications are giving some insurers, armed with the right data,­ new tools to offload much of this work to machines, and mobilize their human talent to tackle higher-level tasks.

After a significant storm or flood, even a streamlined process to handle claims moves station-to-station, from reporting and filing a claim, to physical inspection and investigation, to claim evaluation and approval and payment.

Insurers can now tap into weather observation data and drone-based aerial surveys to automate simpler types of claims payouts in a matter of hours.

Predict future claims

Weather data and machine intelligence also hold promise in long-term risk assessment for insurers handling home or commercial property claims.

Insurance companies can use weather data not just to streamline claims after a storm or hurricane; they can put simulation technology to work to predict future claims and assess portfolio risk on a real-time basis.

If you run a power plant in Michigan, for instance, you can plug in almost every important variable—location, type of weather event, level of severity, time of year, policy information based on location, etc.—and determine how best to quantify the impact, and how to mobilize and prepare call centers and other teams for any scenario.