Lemonade Didn’t Buy Metromile’s Data. It Bought Its Claims Team.

This episode was billed as a quick, off-the-cuff conversation about a single news item — Lemonade’s acquisition of Metromile — and turned into one of the most technically dense breakdowns of insurtech M&A and metric-gaming in the show’s history. Joe Emison, co-founder and CTO of Branch Insurance, joined to work through the deal in close to real time.

In Episode 56 of InsurTechTalk, Joe and I picked apart the two most common explanations floating around for why Lemonade bought Metromile — state licenses and telematics data — and why neither holds up, what Lemonade actually got value from instead, and two specific metrics (loss ratio without loss adjustment expense, and “net retained premium”) that Joe argues investors should treat with real skepticism whenever an insurance startup cites them.

About Joe Emison

Joe Emison is co-founder and CTO of Branch Insurance, a full-stack home and auto insurer built around instant, bundled binding. Alongside building Branch, he’s a frequent voice on insurtech deal analysis and insurance-specific financial metrics.

Two Popular Theories, Neither of Which Holds Up

Joe opened by directly dismantling the two explanations he’d seen circulating for why Lemonade paid roughly $500 million in stock (against Metromile’s own roughly $200 million cash position, implying a net price closer to $300 million) for a carrier writing only about $100 million in premium in force — roughly a 2x premium multiple, a valuation appropriate for a non-growing carrier rather than a growth asset.

The first theory: Lemonade wanted Metromile’s state insurance licenses to accelerate multi-state expansion. Joe was direct that this doesn’t reflect how licensing actually works in practice — owning a carrier’s existing license in a state doesn’t meaningfully fast-track entry if you’re filing even a modestly different program than what that carrier previously had approved; many states will functionally treat that filing as a brand-new entrant regardless of the underlying entity’s licensing history. He also pointed out the going market rate for a single line-of-business license in a single state is roughly $100,000 — nowhere close to justifying a deal of this size on licenses alone.

The second theory: Lemonade wanted Metromile’s telematics and driving data. Joe’s pushback here was sharper. If Metromile genuinely had unique, valuable insight into driving behavior, that should show up in its results — and it doesn’t. Metromile wasn’t profitable and wasn’t growing. Progressive, by contrast, has a mature, filed telematics program and consistently hits its loss ratio targets — and notably, Lemonade’s own newly launched auto product is essentially a direct copy of Progressive’s program, filed as a “me-too” filing (a legally standard practice of adopting a competitor’s already-approved rate filing) in states including Illinois and Tennessee, apparently using Cambridge Mobile Telematics for the underlying scoring. Joe’s pointed question: if Lemonade already had access to a proven, working telematics approach by copying Progressive, why simultaneously acquire a company whose own telematics-driven product never became profitable?

What Lemonade Actually Got: A Real Auto Claims Organization

Joe’s own explanation, and the one he considers most defensible: Metromile’s genuine value to Lemonade is its experienced auto claims organization. Auto claims handling is meaningfully different from the renters, home, and pet claims Lemonade had built its operation around, and Metromile had real, seasoned expertise in that specific discipline — a transferable operational asset in a way that licenses and underperforming data plainly aren’t.

A Puzzling Number: Lemonade’s Renters Loss Ratio

Joe flagged a specific, publicly disclosed number worth real scrutiny: Lemonade’s book is roughly split 50% renters and 50% pet and home combined, and the company has published a loss ratio in the 70s — notably high for renters specifically, where the broader industry average loss ratio runs closer to the 40s. Renters insurance typically has that structurally lower loss ratio partly because retention is naturally low (people move and let a policy lapse rather than actively cancel), which usually keeps claims experience comparatively favorable. Joe was candid that he doesn’t know exactly why Lemonade’s renters loss ratio runs so much higher than the category norm — whether it reflects genuine claims-handling generosity, a disproportionate share of fraud, underpricing, or some combination — but flagged it as a real, unresolved signal.

I offered my own theory in the conversation, framed explicitly as speculation rather than confirmed fact: Lemonade’s widely publicized instant AI claims-approval moment — submit a claim, get an automated approval in seconds, tell your friends — functions as an unusually cheap viral customer acquisition channel. If that’s true, some of what would otherwise show up as marketing spend may effectively be showing up inside the claims and loss cost line instead, which could help explain part of the elevated renters loss ratio.

Why Requiring Telematics Opt-In for Everyone Was a Strange Choice

Beyond the acquisition itself, Joe raised a specific critique of how Lemonade actually launched its own auto product. Progressive’s telematics program — the one Lemonade essentially copied — is deliberately built to serve a very broad range of customers, reaching well into what the industry calls the non-standard segment. Lemonade Car, by contrast, launched requiring every driver on a policy to opt into telematics just to purchase coverage at all. Joe considered this a genuinely strange decision: having gone to the trouble of filing a well-segmented, broadly applicable product, Lemonade then dramatically shrank its own addressable buyer pool right at the top of the funnel — undercutting much of the strategic value of copying a program built specifically to serve the mass market in the first place.

The Real Reason Standalone Auto Carriers Struggle: No Bundle

This was the most structurally important part of the conversation. Joe argued that Metromile’s pay-per-mile pitch — appealing directly to low-mileage drivers who feel overcharged by standard pricing — is a genuinely real but genuinely small niche, and monoline auto insurance is one of the most expensive product categories to advertise, trailing only a handful of categories like mesothelioma-lawyer advertising in cost per click. Targeting only a narrow slice of an already expensive-to-acquire category makes sustainable customer acquisition extremely difficult. He also noted most traditional carriers already discount meaningfully for excess household vehicles beyond what one driver can realistically use at once, quietly undercutting some of pay-per-mile’s differentiation for multi-car households specifically.

The deeper structural point: Joe cited a J.D. Power finding that 78% of homeowners buy home and auto insurance together through a single point of purchase, even if the underlying policies technically sit with different carriers behind the scenes. Bundling isn’t just customer preference — it’s genuinely cheaper to sell, and because home and auto are both largely compulsory purchases (a mortgage requires home coverage; driving legally requires auto coverage), bundled customers churn less, earning a deservedly lower rate. Progressive has referenced this exact dynamic on earnings calls for over fifteen years, using the internal shorthand “the Robinsons” for the archetypal bundled household. Joe’s conclusion: any insurtech with real ambitions of taking meaningful share from an incumbent like State Farm effectively has to build toward genuine home-and-auto bundling (plus adjacent “accommodation” products like umbrella, motorcycle, and boat coverage) — or deliberately choose to own a narrow niche instead of competing for the mass market.

A Better Deal Lemonade Passed On: Noblr

Joe offered a specific alternative path he considered more strategically sound: acquiring Noblr — a well-regarded, technically faithful clone of Progressive’s program — roughly a year before Lemonade ever built its own in-house telematics auto product. Replicating Progressive’s underwriting segmentation is notoriously difficult, which is why, in Joe’s account, most auto insurtechs have avoided trying: Root copied only pieces of Progressive’s approach and (per its own earnings calls) is still actively working to close real segmentation gaps; Clearcover more closely resembles a GEICO-style clone, less segmented than Progressive but still solid; and Loop, a newer entrant, largely builds on Root’s approach with some additional Progressive-style factors layered in. Given how hard a genuine Progressive clone is to build, and that Noblr had reportedly already done it, Joe considered buying that capability directly a stronger move than building a copycat program from scratch and then separately acquiring Metromile for reasons that don’t fully hold together.

On Loop specifically, both of us gave credit for genuinely excellent marketing and storytelling for a young company (an MGA, live in Texas, without published performance numbers yet at the time) — praised as punching well above its weight on brand-building, alongside a nod to Daniel Schreiber’s marketing instincts at Lemonade as a genuine industry standout in that specific discipline.

Why SaaS Metrics Don’t Translate to Insurance

Joe returned to a theme from an earlier conversation between us: the danger of applying software-as-a-service investing intuitions directly to insurance. In SaaS, mispricing a product mostly costs you growth. In insurance, mispricing a product actively loses money every single day that policyholder remains covered — a fundamentally less forgiving failure mode that, in his experience, many generalist venture investors haven’t fully internalized.

Vanity Metric #1: Loss Ratio Without LAE

Joe flagged a specific pattern he considers a real red flag: insurance startups citing a loss ratio that quietly excludes loss adjustment expense (LAE) — the cost of actually processing and adjusting claims — even though a genuine, apples-to-apples comparison against public peers requires including it. The best-run companies land around a 70s combined loss-ratio-plus-LAE figure. Joe pointed to startups citing an already-high-80s loss ratio on its own, before even adding the additional 10-15 points LAE typically contributes — pushing their true combined number over 100%, an unambiguous sign of underpricing — while continuing to cite the incomplete figure in public commentary and comparing it directly against peer numbers that do include LAE.

Vanity Metric #2: “Net Retained Premium”

This was the sharpest technical breakdown in the episode. Joe walked through a clean illustrative example: take 10 policyholders, each paying $1,000, for $10,000 in total premium. At renewal, one leaves — a straightforward 90% policy retention rate, which is how most insurance companies traditionally report retention. But if the carrier simultaneously raises rates by 12% on the remaining nine policyholders, total retained premium dollars can actually come out to roughly 102% of the prior year’s total — even though real customer retention was only 90%. Joe specifically named this as Hippo’s disclosed metric, “net retained premium,” and noted that raising your price on the same product improving your own reported “retention” number is not something any policy-based retention metric would ever show, and that comparing this premium-dollar-based figure directly against a genuinely policy-based retention number (like Progressive’s) is an apples-to-oranges comparison that’s gone largely unchallenged in press coverage and analysis.

He traced the metric’s origin to a genuinely useful concept in software: net negative churn, where an actively engaged, growing SaaS customer naturally spends more over time by adding seats or upgrading tiers — a real signal of expanding value delivered. That concept doesn’t transfer cleanly to a commoditized product like home or auto insurance, where a renewing policyholder paying more is usually just inflation, rising replacement costs, or a straightforward rate increase, not evidence of a deepening customer relationship. As a real-world contrast, Joe cited Allstate, which has been net-shedding actual policyholders for roughly 30 years while its total premium in force has still grown modestly overall — a legitimate, disciplined outcome specifically because Allstate doesn’t try to frame that premium growth as a substitute for genuine retention health the way some insurtechs do with premium-based metrics.

The Two Diligence Questions Worth Asking

Joe distilled his framework for evaluating any insurance startup into two direct questions. First: what loss ratio, inclusive of LAE, did the company actually target, and what did it actually hit — and if there’s a meaningful gap, why. Second, raised directly by the Metromile deal itself: what is the real, specific addressable market for this product, since almost every insurance startup pitch deck cites an enormous total addressable market that rarely reflects the narrower niche the actual filed product is built to serve. A related, harder question for any investor to resolve: is this fundamentally a technology company that happens to sell insurance, or an insurance company that happens to have excellent technology — since those two framings imply genuinely different, appropriate valuation multiples.

A Footnote on Stock-Deal Mechanics

Because the Metromile acquisition was structured as a stock-for-stock deal rather than an all-cash transaction, its real value kept moving with Lemonade’s own share price throughout the months it took to close — meaning the widely cited headline figure (roughly $500 million) was already somewhat stale by the time of this recording. Both of us noted this as a real, underappreciated risk in stock-financed M&A generally: a long closing period exposed to market volatility can lead to deals being materially renegotiated, or falling apart entirely, if the acquirer’s own stock moves unfavorably before close.

A Maturing Investor Base — With a Real Caveat

Closing the conversation, we discussed the insurtech investor community roughly six to seven years into its own maturation, with a recognizable second generation emerging (former principals and associates now raising their own funds). Both of us were careful to draw a clear distinction here: dedicated insurance-specialist investors, in our shared experience, generally do understand loss ratios and target-setting well and make genuinely sophisticated underwriting-adjacent decisions. The real blind spot sits with generalist venture capitalists less familiar with insurance-specific metric-gaming patterns, and with retail and public-market investors who have comparatively less experience parsing insurance disclosure norms and non-GAAP conventions than they might with, say, a typical SaaS company’s reporting.

Key Takeaways

• The two popular explanations for Lemonade’s Metromile acquisition — state licenses and telematics data — don’t hold up under scrutiny: licenses don’t meaningfully accelerate multi-state entry, and Metromile’s own unprofitable, non-growing results undercut the idea that its data translated into any real underwriting edge
• The most defensible source of real value in the deal is Metromile’s experienced auto claims organization, a genuinely different discipline from the renters, home, and pet claims Lemonade had built around
• Lemonade’s disclosed renters loss ratio (70s) runs notably higher than the renters industry norm (40s), an unresolved signal worth watching regardless of the specific cause
• Requiring every driver on a policy to opt into telematics before purchase, despite filing an otherwise broadly segmented Progressive-style product, needlessly shrinks Lemonade Car’s addressable buyer pool at the very top of the funnel
• Real, durable growth in personal lines requires genuine home-and-auto bundling — 78% of homeowners already buy both together through a single point of purchase — making bundling, not standalone product cleverness, the actual structural key to displacing incumbents like State Farm
• Two specific metrics deserve real skepticism from insurance investors: a loss ratio quietly reported without loss adjustment expense, and “net retained premium” style figures that can rise purely from rate increases even as genuine policyholder retention declines
• The best diligence questions for any insurance startup are refreshingly simple: what loss ratio (including LAE) did you target versus actually hit, and what is the real, specific addressable market for this product — not the generic total market size every pitch deck cites