The insurance industry has long grappled with the catastrophic damage brought by extreme hail events. Crops, roofs, vehicles, and many other forms of capital are threatened, with between 8 and 14 billion in losses experienced every year in the United States. The conventional approach by insurers has been a forensic one: looking for a specific event which caused the damage, assessing a submitted claim, and responding to hail risk after the damage has been done. Complicating matters further, data from the largest hail storms, the ones most looked at by insurers, does not fully account for the losses claimed by policyholders. The industry’s common explanation for this is the presence of rampant fraud. Experimental evidence from IBHS, where common roofing materials are exposed to repeated impacts from simulated hailstones, shows that while widespread fraud is possible, a large amount of damage also comes from accumulated small stone exposure. Fortunately, this explanation provides more actionable strategies for insurers wishing to address hail risk in their portfolio.
ZestyAI data scientist Bryn Ronalds, PhD, was invited to present new research at the annual IBHS hail conference hosted at the National Center for Atmospheric Research in Boulder, Colorado to explain what a fresh perspective on hail offers the insurance industry. The panel, which balanced private industry scientists with academic researchers, provided compelling reasoning for changing how hail events are tied to actual damage. Ronalds leads the effort at ZestyAI to rethink how insurers can make use of real loss data, aerial imagery, computer vision models, property-specific insights, and climate science to make the best estimate of hail risk before the damage occurs. What used to be a forensic process, focused on matching hail events to observed losses, is quickly becoming a proactive risk decision informed by cutting-edge science. Insurers now have access to predictive climate risk models that assess all the available data to create a highly accurate risk score for each individual property.
Insurers have historically focused on large hail events, often limited to reports of hailstones exceeding two-inches, because the data is readily available and the common understanding in the industry is that these storms cause the most damage. Forthcoming cutting-edge research from both IBHS and ZestyAI, previewed by Ronalds at the conference, shows that smaller hail event data is better able to explain hail claims than the large hailstone data. While it is true that one milder hail event may not cause more damage than one record-setting storm, smaller events occur so much more frequently that they are cumulatively more responsible for losses.
Thanks to improvements in data collection, research into mechanisms driving hail damage, and advancements in AI-powered climate risk modeling, ZestyAI is prepared to help insurers make sense of the threat of hail damage. ZestyAI created Z-HAIL™ by leveraging the success of Z-FIRE™, which gives insurers and property owners in wildfire-prone areas across the US better visibility into their risk exposure. Z-HAIL™ is the first risk solution that combines individual property features across North America, material and climate science, and state-of-the-art machine learning methods to help insurance partners assess both the frequency and severity of claims caused by hail damage. With this risk solution insurers are able respond to hail risk using AI models built with cutting-edge science from organizations such as IBHS. Z-HAIL™ also allows insurers to more accurately price risk and provide greater transparency to their customers. To learn more about Z-HAIL™, contact us at firstname.lastname@example.org.