How AI can mitigate roof-related insurance losses
Incorrect roof age data is a significant concern for P&C insurers, costing $1.3 billion per year in premium leakage, according to CBIZ.
More than half of insurance claims involve roof damage, according to Lending Tree, and these losses are often caused by wind, hail and weather-related water damage. The Insurance Information Institute (Triple-I) reports that 39.4% of homeowners insurance losses in 2021 involved wind and hail, while 23.5% were due to water damage and freezing. It’s fair to say roof repairs cost property and casualty insurers a pretty penny annually, but companies such as Nearmap and ZestyAI are working to mitigate these losses with more accurate data on roof age for carriers.
“The predominant way that carriers get that [roof age] information is a homeowner-supplied roof age,” said Kumar Dhuvur, co-founder and head of product at ZestyAI, makers of Digital Roof™. “We polled a random sample of homeowners across the country, and we found that more than two-thirds of them did not know what the age of their roof was, especially if they were not the owner when the roof was replaced.”
Roof age accuracy
Carriers may restrict coverage on properties with roofs that haven’t been replaced for 20 years or longer or may deny coverage altogether. Incorrect roof age data is a significant concern for P&C insurers, costing $1.3 billion per year in premium leakage, according to CBIZ. Homeowners knowingly or unknowingly providing false roof age information may initially benefit from lower insurance premiums but ultimately have a roof-related claim denied due to the deception. Insurers need a more accurate way to determine roof age.
P&C carriers may also acquire roof age data by accessing building permits. Still, not all towns require homeowners to have a building permit for their roof, especially after a severe weather storm, says Dhuvur. Other traditional methods involve in-person inspections, but aerial imagery can help narrow down the precise year the roof was replaced.
“When building permits are not available, we rely on historical aerial imagery to be able to figure out when a roof was replaced, and the combination of this essentially makes this product really accurate. You’re able to deliver accuracy that’s about 90%,” said Dhuvur.
More accurate and robust data about roof age benefits insurers. Carriers with this information can provide the appropriate coverage and adjust the premium to account for the true roof age, ensuring they pay the correct amount for a one-year-old roof replacement versus a 10-year-old roof.
Aerial imagery
Insurers have long used aerial imagery to calculate roof age, but that requires human agents to look over years of images to pinpoint when the roof was replaced. This also requires ample time and resources, but using AI to pour through this data offers a great advantage.
“We’re taking aerial imagery. We’re processing it with AI… The models we present, we try not to do black box models. We call them glass box models because they’re transparent, right? You can see if we say your roof score is X. We’ll show you what the computer saw that said that. Missing shingles here, or water ponding, or whatever the malady might be… it has to be explainable,” said David Tobias, general manager of insurance at Nearmap.
Policyholder benefits
Roofs become more vulnerable with time as the glue and binding of the shingles deteriorate and the substrate weakens. Homeowners with new roofs reap benefits on their premiums and at claim time. The Texas Department of Insurance reports the average payout for a five-year-old roof is $4,500, $3,000 for a 10-year-old roof and $0 for a roof 20 years old. Policygenius suggests homeowners with new roofs could save $151-$425 on their annual premiums compared to those with older roofs. Policyholders with new roofs may also have more access to insurance coverage and are less likely to have restricted coverage or face nonrenewal. Demonstrating these cost savings may incentivize insureds to maintain and repair their roofs before severe weather hits their area.
“If carriers now have accurate information, they’re able to accept policies with precision,” says Dhuvur. “They don’t have to make blanket decisions about whether they want to write policies in an area. They can make individual decisions… Instead of average premiums, you will get premiums where properties with better risk are going to get lower premiums.”
Related: