Why many insurtech car insurers struggled with losses

Most of the first wave of digital car insurers lost money because they sold policies faster than they learned what those policies would cost, then ran into the steepest jump in repair and injury costs in decades while paying for advertising and reinsurance. Root, the best documented case, reported a net loss every year from 2018 through 2023 and net income of $30.9 million in 2024 and $40.3 million in 2025, according to its filings. Metromile was absorbed by Lemonade in 2022, and Lemonade was still reporting net losses in 2025.

This is a piece about mechanisms, not about whether any one company is a good or bad choice. Each of the forces below hits every young insurer, and each is visible in public numbers. The figures here come from annual reports the companies filed with the SEC, plus AM Best's industry data as reported in the trade press. They are dated, and they describe the past; a company that lost money in 2021 can be profitable now, and Root's own recent results show it.

What a loss means here

"Struggled with losses" covers two different things, and mixing them up causes most of the confusion.

An insurer's loss ratio is claims paid and reserved (plus the cost of adjusting them) divided by premium earned. If it is 80%, the insurer spends 80 cents of each premium dollar on claims. The expense ratio is everything else it spends running the business: advertising, commissions, salaries, software, premium taxes. Add them and you get the combined ratio. Above 100%, the insurer loses money on underwriting before counting what it earns by investing premium in the meantime. Our companion piece on how to read an insurer's combined ratio and surplus walks through reading the numbers.

A young insurer can fail either test. It can pay too much in claims relative to premium, or it can spend too much to get the customer. Root's records show both at once: in 2021 its gross loss ratio was 86.0% and its net expense ratio was 134.5%, which produced a net combined ratio of 260.9%, according to its 2022 annual report. Put plainly, for each dollar of premium it kept after reinsurance, it spent about two dollars and sixty cents.

The record, in dated numbers

Root's annual reports give a continuous series. All figures are from the company's own filings: the 2020 report, the 2022 report above and the 2025 report.

Two things stand out. First, the gross loss ratio, which measures claims against premium before reinsurance, moved only slightly between 2020 and 2022, from about 82% to 86% and back to 82%, while net losses swung widely. The big swing is in expenses and in how much of the business the company kept. Second, the turnaround came with a large fall in the loss ratio (82.1% to 58.0% between 2022 and 2025) and in the expense ratio. Root's report attributes the profit trend to rate actions and better segmentation; the numbers show it happened alongside premium per policy rising from $939 in 2020 to $1,220 in 2022, about 30%.

Lemonade is a multi-line insurer, so its numbers mix home, renters, pet and car. Its 2025 annual report shows a net loss of $165.5 million in 2025 and $202.2 million in 2024, and a gross loss ratio of 64% in 2025 against 73% in 2024. Metromile, a pay-per-mile car insurer that went public through a merger with a special purpose acquisition company on February 9, 2021, reported a net loss of $216.5 million for 2021 in its annual report, up from $120.1 million in 2020. Carrier Management reported on July 29, 2022 that Lemonade had closed its all-stock purchase of Metromile, which it announced in November 2021, and that Metromile brought $155 million in cash, over $110 million in premiums and an insurer licensed in 49 states.

Growth versus loss ratio

New customers lose more money than old ones. Root said so in its own 2020 report: renewal premiums, those from a customer's second term onward, "have lower loss ratios as compared to new premiums in the customer's first term." The reasons are structural. At the first term an insurer knows the least about a driver, and it has paid for advertising, a quote and a policy issue. By the second term the driver has a claim history with that insurer, and the acquisition cost is behind it.

For a company growing fast, most of the book is first-term business, so the loss ratio and expense ratio look worse than they will once the book matures. Insurance accounting adds a second squeeze: regulators require capital relative to premium written, while the premium is earned over the policy term, so growth consumes capital before it produces profit. Hence the "growth versus profit" tension that companies describe in their filings.

Root's 2022 numbers show the reverse. Policies in force fell from 354,371 at the end of 2021 to 220,354 at the end of 2022. The report attributes the lower premium earned to "reduced marketing expenditures," partly offset by rate increases. Cutting growth cut the first-term mix and the acquisition cost, and the net expense ratio dropped from 134.5% in 2021 to 72.2% in 2022 and 27.9% in 2024.

Adverse selection: when cheap attracts the wrong risk

Adverse selection is what happens when the customers most likely to claim are the ones your price attracts most strongly. It is a well-known hazard for any new entrant that prices below the market in some segment.

Take a hypothetical example. Say 1,000 drivers are shopping, and half are low-risk, expected to cost an insurer $1,000 a year in claims, while half are high-risk, expected to cost $2,000. An established insurer with many years of data charges about $1,100 to the first group and about $2,100 to the second. A new insurer's model cannot tell the groups apart as well, and it quotes both at $1,300.

The low-risk drivers see $1,300 against $1,100 elsewhere and stay put. The high-risk drivers see $1,300 against $2,100 and switch in droves. The new insurer's book is mostly high-risk: $1,300 premium against $2,000 expected claims is a loss ratio of about 154% before any expenses. Nothing about the drivers' behavior was irrational; the insurer's pricing was most attractive precisely where it was most wrong.

This does not mean a data-driven model cannot work. It means the first year of a new model's pricing is a test, and the test results arrive as claims, with a lag. A telematics score can be a better predictor than age or ZIP code, and the main insurers also use telematics. AM Best's Edin Imsirovic made the competitive point in an October 2021 AM Best article: learning tech alone will not make insurtechs profitable, he said, because incumbents already use technology more heavily than the industry overall. Better data helps only to the extent it prices a risk better than the competitors' data does.

Claims inflation after 2021

The mechanism above is hard on any new entrant in a steady market. The market after 2021 was not steady. AM Best's December 2023 market segment report, as reported by Reinsurance News, found that U.S. personal auto insurers posted a 112.2 net combined ratio in 2022, nearly 11 points worse than 2021 and about 10 points worse than the 10-year average and median, a $33.1 billion underwriting loss. The report cited economic inflation, supply chain disruptions and vehicle technology as drivers of higher claim costs, plus higher accident frequency.

Why does a surge in claim costs hit a young insurer harder than an old one?

1. Rates lag costs. Insurers set a rate for the year ahead from past claims. If repair and medical costs rise during the year, the rate is too low until the next filing is approved and takes effect. In states with prior-approval rules the delay can run for months; see loss ratios and rate filings: how insurers justify an increase, and our article on Profitability by line and state: what NAIC data says about margins for how margins differ by state.
2. There is no cushion. A large incumbent has decades of retained earnings and a diversified book. A young insurer has the capital it raised, and every point of underwriting loss draws it down.
3. Reserves are estimates. Claims, especially injury claims, settle years after the crash. A young insurer has little history of its own to calibrate reserves against, so an early estimate can prove too low and be raised later.

Root's report is candid that claims inflation is a risk: its 2022 report warns that changes in bodily injury claim severity are "impacted by inflation in medical costs, litigation trends" and other factors. These are not unique to insurtechs. They matter more to a company with a thin balance sheet.

Reinsurance and funding: the price of growing on borrowed capital

Because regulators require capital in proportion to premium written, a fast-growing insurer without much capital can buy reinsurance to pass along part of the risk and part of the premium. Root's 2022 report says it ceded about 56.9% of its gross premiums earned to third-party reinsurers in 2021 and 55.6% in 2022. That keeps a small company within its capital limits, but the company gives up much of the premium, and so much of the profit if it turns out the business is profitable. Root's 2025 report describes the later change: as its gross loss ratios stabilized, it "strategically reduced" its use of external quota share. Its ceded premiums earned fell from $160.1 million in 2024 to $64.3 million in 2025 against gross premiums earned of $1,466.0 million.

Reinsurance is also a lever that cuts both ways. Metromile's 2021 report shows net premiums earned rising from $12.5 million in 2020 to $75.6 million in 2021, mostly because it ceded $51.7 million less premium to reinsurance partners. Retained premium looks like growth, and it is also retained risk.

Funding is the other side. AM Best's October 2021 article describes a "continued record-breaking flow of funding" into insurtech, noting that Root went public in October 2020 and Metromile in February 2021. Root's 2020 filing says its initial public offering and concurrent private placements brought net proceeds of $1.1 billion. In a funding wave, the capital arrives before the evidence about pricing does. That makes growth affordable, and it also makes bad pricing affordable for longer.

Consolidation: dated examples

The simplest summary of the outcome is who ended up where. As of the dates in the filings:

• Metromile into Lemonade (closed July 2022). Lemonade bought the company, kept the Metromile insurance entity as a subsidiary, and its 2025 annual report still lists Metromile Insurance Company among its insurers. The Metromile brand is being phased out, per our Metromile profile.
• Root's retrenchment (2022). Policies in force fell by about 38% in a year (from 354,371 to 220,354) as marketing was cut, before the company returned to growth through partnerships with other sellers.
• Lemonade's expansion into car (2021 onward). Lemonade launched Lemonade Car in Illinois in 2021, a week before announcing the Metromile deal, and per its 2025 report still had a net loss that year.
• Hybrid models. Clearcover's site lists its own carriers in Illinois and other carriers' paper in Texas and California as of November 2026. Our article Full-stack insurtechs vs MGA models: who carries the risk covers why that matters.

None of these is a failure story in the sense of unpaid claims. In each case policies stayed in force and claims continued to be paid. Consolidation matters to you mainly because a brand, a carrier or a claims process can change under a policy.

What a policyholder should look for in a young insurer

The goal is not to avoid new insurers. It is to know what you are relying on. Use this checklist.

Ratings need a note. Lemonade's 2025 annual report states that its insurance subsidiary holds an "A" (Exceptional) Financial Stability Rating from Demotech, Inc., and that it does not have a rating from A.M. Best. That says the two agencies' scales are not interchangeable, and it says nothing about whether the company will pay your claim. Our explainer AM Best financial strength ratings explained describes the differences.

The guaranty fund point is worth knowing. The National Conference of Insurance Guaranty Funds says that state property and casualty guaranty associations cover only licensed insurers, that they typically step in after a court issues an order of liquidation, and that the most common cap per claim is $300,000, though it varies by state. The details are in the NCIGF's insolvency overview and in our guide on what happens if your car insurer goes insolvent. The NAIC's Consumer Insurance Search lets you look up a company and its subsidiaries, and it advises checking your state insurance department to confirm a company is licensed.

A worked example of how losses build, with made-up numbers

Say a young insurer writes $10 million of premium over a year. These numbers are invented.

• Claims and claim handling cost 95% of premium: $9.5 million.
• Advertising, commissions, salaries, software and taxes cost 45%: $4.5 million.
• Combined ratio: 140%. Underwriting loss: $4 million.

Investment income on the premium might offset a little. The rest comes out of the capital the company raised. If it raised $20 million, it is a fifth gone. Now suppose it grows 50% the next year at the same ratios: the loss is $6 million, and the regulator's capital requirement has also risen, because there is more premium. The company has to cut growth, raise rates or buy more reinsurance, and each choice has a cost. That is the loop Root's numbers show between 2020 and 2022, and the way out is the one its numbers show afterward: lower loss ratio, lower expense ratio, less ceded premium.

Common questions

Did these insurers fail to pay claims?

The filings above do not describe that. Root, Lemonade and Metromile continued to operate as licensed insurers during the periods discussed, and Metromile's business became part of Lemonade. The losses were the companies' and their investors', funded by the capital raised. The risk to a policyholder is indirect: a rate increase, a nonrenewal, a changed claims process or a brand change.

Are digital insurers still losing money?

It depends on the company and the year. Root reported net income in 2024 and 2025 and a net combined ratio under 100% both years. Lemonade reported net losses in both years, with its gross loss ratio improving from 73% to 64%. Check the latest filing for any company you are considering.

Does a low price mean a young insurer is in trouble?

Not by itself. A low price can reflect a good fit with your risk profile or a promotion. A price far below every other quote for the same coverage is worth a second look. The new customer is the one the insurer knows least about, and the price may rise at renewal; compare it with what other insurers charge for the same limits.

Is it risky to buy from a company that is not rated by AM Best?

It means one particular agency has not rated it. Some companies hold ratings from other agencies. Look at which agency issued the rating and how long the company has operated, and check that the carrier is licensed in your state.

How do I know whether my insurer is the one filing the losses?

Public companies file annual reports at the SEC. The insurer that issues your policy may be one of several subsidiaries, so find the legal name on your declarations page, then check the parent's report for the group that includes it.

Key takeaways

• Young car insurers lose money through two channels, claims that exceed premium and the cost of acquiring customers; Root's 2021 net combined ratio of 260.9% shows both at once.
• Growth makes things look worse before better: first-term business has higher loss ratios, and capital rules tie up funds as premium grows.
• Cheap pricing can attract the riskiest customers (adverse selection), and the 2022 industry combined ratio of 112.2 (AM Best, via Reinsurance News) left little room for error.
• Reinsurance and venture funding allow rapid growth but cost margin; Root ceded about 56% of its premiums in 2021 and 2022 and about 4% in 2025.
• Check the licensed carrier, its rating and who issued it, how long it has written car insurance in your state and its complaint record, as of the current month rather than the last headline.