A generation of startups promised to transform the insurance industry with new types of data and algorithms to more accurately assess risk and price policies nearly a decade ago. So far, this has not occurred.
Lemonade, Root Insurance, and Hippo, three high-profile insurtechs that have since gone public, have each lost tens of millions of dollars in their most recent quarters and seen their share prices plummet in recent years.
At first, the insurance pricing process, which is heavily reliant on algorithms and mathematical modeling, appeared ripe for disruption due to advances in the sheer amount and variety of data digitally-native companies could suddenly collect on customers.
However, the Silicon Valley axiom of “move fast and break things” hasn’t been enough to transform an industry founded on centuries of observed human behavior, massive marketing budgets, and a keen understanding of the regulatory environment.Founded in 2015, Lemonade initially aimed to sell renters and homeowners insurance. It was worth $9.87 billion at its peak in 2021; it’s now worth $1.23 billion. Root Insurance, also founded in 2015, began with the idea of using telematics—or in-car data—to offer personalized auto insurance based on how people drive. In 2020, it was worth roughly $6.8 billion, and has since swooned to about $67 million. Property and casualty insurance startup Hippo went public at a $5 billion valuation in 2021. It is now worth around $425 million.
So far, the insurtechs have been slow to gather and contextualize enough data to actually build better models. Regulations have restricted the use of some of their data and differentiated pricing. And it has been difficult to chip away market share from established industry giants.
“For those who believed they could build a full-stack insurance company that would become a market-leading, at-scale carrier—I think that thesis has not materialized,” said Tanguy Catlin, a senior partner at McKinsey.
All three upstarts say profitability is on the horizon, some blaming their recent poor showings on inflation and a pullback in tech investment over the last year. Their models are becoming increasingly accurate, they say, as they gather more data and learn how to contextualize it, and losses are trending downward. They also said they’ve had a meaningful impact on the customer experience of buying insurance, which can now be done online.
But as far as transforming the risk analysis and pricing of insurance, that remains to be seen.
“To date, we’ve not really seen disruption,” said Jefferies analyst Yaron Kinar.
“We did a fairly shoddy job of pricing and identifying risks. And we knew we would,” said Lemonade co-CEO and co-founder Daniel Schreiber of the early days of the company. The problem was that Lemonade, along with other insurtechs, started without access to the historical data sets that incumbents have been gathering, in some cases, over many decades.
To build those more nuanced and complex sets of data takes time, Schreiber said. It isn’t just a matter of collecting enough data points, but also of identifying those that signify risk, which only happens once a claim is filed, he added.
It’s a similar story at auto insurance startup Root, which built its business around the use of telematics. Through a smartphone app, Root could track the driving styles of users for two weeks, then offer them a personalized price based on how they drove. But Root didn’t know for sure whether certain driver behaviors actually correlated to higher risk until accidents occurred and the claims started coming in, said Chief Executive and Co-founder Alex Timm.
Root grew fast, and in its early years had tens of thousands of claims coming in to work with. Once it was able to refine its model, the company went back and adjusted prices, Timm said.
But that was a risky play, said Jefferies analyst Kinar. The type of clients who gravitated to insurtechs were younger and more price-sensitive—meaning they were more likely to leave the platform if they saw a meaningful rate increase.
Root said 60% of its customer base is under 30.
Chevy vs. Ferrari
Although startups had the digital-forward advantage of being able to collect tons more data points from smartphones, online interactions and other sources, regulations prevented them from actually using much of it for risk analysis and pricing.
Lemonade says it collects hundreds of data points on people who visit its website to buy insurance. The company can see whether someone came to the website by searching for “cheap insurance” versus “good insurance,” as well as other information, Schreiber said. Lemonade can correlate some of these online behaviors to more risky individuals.
That provides a fuller picture of a potential customer than having them simply answer 20 questions over the phone, the way they might in traditional insurance, he said. But regulation stops Lemonade from using information about observed behavior for pricing, Schreiber said. The company still uses it for targeted marketing, he added.
Regulations vary state by state on what data can be used for pricing. The use of telematics to price auto insurance is permitted in some, but not others, said Timm. For that reason, Root cannot use telematics to price policies in California.
At Hippo, the company can link to its customers’ smart home devices, but no data from those devices can be used to assess risk. The company can, however, use the fact that a given device is turned on to give the customer a discount.
When insurtech was hot, startups were pitching all types of new data, including voice analysis or weather predictions five years into the future, as ways of transforming risk analysis, according to Michel Leonard, chief economist and data scientist at the Insurance Information Institute. But those lofty ideas were dashed on the realities of regulation, he said.
“At first, it’s a great sell, great technology,” Leonard said. “But then when you come to underwriting, well, what’s the point? You can’t use it and you don’t have legacy data, so you’re really at a disadvantage.”
Given those restrictions, the startups’ souped-up computing turned out to be superfluous, he said, like building a “Ferrari, when a Chevy is good.”
As they worked to provide more accurate pricing models, the startups found themselves in an intensely competitive industry. The cost of acquiring customers is enormously high in the insurance business, said McKinsey’s Catlin. In direct-to-consumer auto insurance, for instance, some market leaders spend billions a year on marketing, he said.
“You’re coming up against two—900-pound gorillas is an understatement,” Jefferies’ Kinar said referring to Geico and Progressive. The incumbent advantage, he said, is both in their scads of marketing dollars as well as their deeper wells of data to figure out how to put those dollars to best use.
“It’s certainly not for the faint of heart,” commented Timm.
Analysts agree that while insurtechs have yet to transform the industry, they have left their mark on the way providers interact with customers, with faster, digital channels.
Entrenched companies are working to catch up. At Allstate, the amount of time it takes to buy a policy has dropped significantly, said Suren Gupta, president of enterprise solutions. A process that might have taken around 20 minutes can now be done, in some cases, in less than five, he said, adding that some of the changes he’s put into motion have been inspired by insurtechs.
Indeed, the greatest value of insurtech startups might be as incubators for the industry, Catlin said.
“You put money into insurtech to develop cool technologies, and then the traditional carriers find a way to access that technology, whether they buy the insurtech or they license the technology,” said Catlin. “That’s where I’ve seen most of the evolution.”
Lemonade, Hippo and Root said their virtual origins still give them a head start over industry powerhouses working from legacy systems. Hippo also said its agility gives it the opportunity to adapt to market conditions better by filing rate changes faster and more frequently than its more established rivals.
As far as promises around industry “disruption” go, Kinar said, “these days, that word doesn’t come up much at all anymore.”