Digital insurance platform Lemonade said artificial intelligence helped it improve its loss ratio.
The figure, the ratio of losses to premiums earned, improved by 12 points year over year, the company said in a Tuesday (Feb. 27) letter to shareholders announcing quarterly earnings results.
“Our improving loss ratio, and our growing powers of prognostication, owe much to our suite of homegrown AIs, and specifically to one we refer to fondly as ‘LTV,’” Lemonade said in the letter. “This composite AI harnesses some 50 machine learning models to predict, for each customer, their lifetime value, or LTV, in net present dollar terms.”
These models, the letter said, make a range of real-time projections, including the lifetime loss ratio of each customer, although their main output is a simple dollar figure that represents the “net present value” of a given customer.
“Many of our business processes rely on these few digits,” the letter continued. “For example, LTV dictates which products, which geographies and which advertising campaigns receive the incremental marketing dollar, a hugely influential decision with real impact on our loss ratio and profitability.”
Lemonade added in the letter that its loss ratio will likely continue to trend downward but warned that “seasonal headwinds” could persist in the next few quarters.
The letter noted a variety of factors that can affect loss ratios from season to season, like an uptick in water damage claims from burst pipes in the winter or wildfire season on the West Coast during the summer.
Lemonade is “AI native,” Chief Financial Officer Tim Bixby told PYMNTS in an interview late last year, adding that the company’s “real advantage lies in the depth of data that we collect about each customer when we onboard them and when they get a quote and a policy.”
PYMNTS also examined the role of AI in the insurance world in an interview with Foxquilt CEO Mark Morissette, whose company uses data analytics and AI to recommend the best insurance coverage and price through brokers, agents and enterprise partners.
“You need to define that customer and then go and solve their problems,” he said. “But at the same time, you need to generate a sustainable business model that generates underwriting profitability and build a smarter machine that facilitates solving customer problems.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.