In Case You Missed it:

Lemonade offers renters, home, auto, pet & life insurance in the USA and parts of Europe. The company’s tech-and-AI-native foundation cuts out hefty fixed costs associated with legacy models to lower cost of service. This enables it to routinely outcompete substitutes on price. It builds on this edge by obsessively iterating on its underwriting algorithms to ensure accurate underwriting. This has enabled accelerating top-line growth, sizable fixed cost leverage and brisk margin improvement for years.

For a detailed review of its latest investor day, where I dig a lot more into what makes Lemonade special, click here.

Note that a lower rate of ceded premiums to reinsurers is currently propping up revenue growth. Lemonade’s rising confidence in their underwriting, profitability and liquidity all gave it the confidence to keep a larger portion of their overall demand. This means more revenue on the same base of premiums and, if pricing risk is well executed, should be a profit dollar accelerant as well. IFP and GEP growth both were strong and offer a better sense of demand right now, as both exclude that noise.  This is the 8th consecutive quarter of accelerating IFP growth and marks Lemonade reaching its 30% Y/Y growth acceleration target a full two quarters early.

I’d like to highlight the third demand chart for a moment. You’ll notice that the large headwind from Lemonade cutting negative profit home plans in California continues to greatly weigh on that product’s expansion. The headwind actually got worse this quarter vs. Q1 or Q2, yet they meaningfully accelerated premiums growth. It’s exciting to think that this product will again be unleashed at some point and the growth engine already looks this good before that even happens. One more note – car insurance has fully arrived. 40% Y/Y growth for that product marks a sharp acceleration.

“All products and regions contributed to the dynamic of accelerating top line and improving profitability.” – Co-Founder/CO-CEO Daniel Schreiber

The rest of this full review & the Duolingo review are for paid readers. They include overviews of financials, guidance, valuation, conference calls, other investor materials and my overall takes. It's detailed yet condensed work to save you time. Subscribe below to read that and 40+ earnings reviews every season.

Learn more

On gross loss ratio (GLR), prior period development (updated estimate of losses from recent cohorts) boosted the Y/Y strength. Of the 11 points of improvement, 2 points were related to that. Still excellent.

Free cash flow (FCF) rose from $5M to $18M Y/Y. This is greatly helped by its synthetic agent financing arrangement with General Catalyst. That partner finances most of LMND’s growth spend, so that cost isn’t a cash drain like it is for a typical company. This is why I think it’s important to focus on income statement profitability trends, which look great.

“We're well positioned to fund our growth strategy without need for additional capital.” – CFO Tim Bixby

AI-Powered Efficiency Edge Enabling Differentiation:

Another quarter of overall OpEx growing well below the pace of revenue. Specifically, OpEx rose by 13% Y/Y, which is a bit faster than recent quarters, but they don’t view that as a new theme. Leadership, based on their guidance, expects cost growth to fall below 10% Y/Y again next quarter while IFP growth remains around 30% Y/Y. More leverage is coming. When digging into the sources of OpEx growth, the patterns are identical vs. recent quarters. They’re just spending more on growth… but even that ramp is quite reasonable. Growth was just 16% Y/Y – again at roughly half the pace of top-line expansion. They now see growth spending for the full year coming in at $180M vs. $170M previously because they’re so encouraged by what they’re seeing and are taking advantage of highly productive growth opportunities. They are not wasting dollars and not getting less productive in customer acquisition. Lifetime value to customer acquisition cost (LTV/CAC) is a popular metric for gauging how much you’re spending to add a new customer and how much value they’re delivering you. And? They are maintaining a strong 3x ratio in that regard.

As we’ll see later on in this piece, rapid growth in other expense buckets is just not needed anymore. Lemonade’s AI-native core is rapidly finding new ways to adopt and more quickly deploy novel technologies to extract more efficiency from their existing assets. Loco, its no-code insurance application builder, is a good example of this. It allows LMND employees to vastly accelerate software build time and improve customer service.  And the firm’s strict product update testing ensures they know whether this technology will be a winner or not before they fully deploy it.

This is why headcount is flat over the last 2 years and how they expect to keep delivering 30% demand growth with more single-digit cost growth going forward. G&A growth may stay around the 12% Y/Y rate we saw this quarter, but that’s because interest expense is rising in connection with their General Catalyst growth spend financing arrangement. It’s a lagging byproduct of growth spend. All of this leaves us with a recipe for a near-term profit inflection… and a subsequent explosion. Every sign points to that being inevitable at this time. And management remains highly confident in reaching EBITDA positive by Q4 of next year.

Loss Ratio Tailwinds & LAE:

The quarter also convincingly represents record lows for GLR and the less volatile trailing 12-month GLR. This paved the way for fantastic GPM leverage. As the book scales, underwriting improves and the new business penalty diminishes, loss ratios are brightening. And there’s another exciting item to discuss that leadership spent more time on this quarter. That is Loss Adjustment Expense (LAE) leverage. LAE includes all costs to handle and resolve claims (not the actual claims). It’s a great measure of how efficiently a company can handle routine customer inquiries. Despite its smaller scale, Lemonade is showing a clear ability to operate more efficiently than its competition. Specifically, LAE for Lemonade is now 7% vs. 13% just 3 years ago. That 7% compares to 9% for massive incumbents, which is especially impressive considering LAE benefits from more scale. LMND is enjoying this positive pattern for all of its products, showing us that the improvements are structural rather than mix-related. For these reasons, Lemonade sees LAE falling to an incredible 3.5% over the next couple of years while the business doubles. 

Where the Cost Edge Comes From & Why it Matters? 

Lemonade has flat-lined non-growth spend because it has been so successful in using AI to minimize variable costs as it grows. Whether that’s customer service, claims handling or anything else, they’re not reliant on labor growth to support a larger business like their incumbents are. This should mean their long-term margin and operating efficiency ceilings are higher than the field. 

They’re not victims of a century of disparate integrations and point solutions needing to somehow be jumbled together. They’re not built on an archaic data architecture that inhibits lucrative customer insight. They’re built for the modern world and constructed to embrace change more quickly than the competition. The data pipelines are crafted with unstructured, jumbled chaos in mind, allowing LMND to ingest and make sense of whatever it needed in an interoperable manner. They don’t move quickly in random or accidental ways. They do so in entirely data-driven ways, with light-weight, world-class products and APIs making sure these insights can rapidly turn into better outcomes.

Why does this relative efficiency edge matter so much? Insurance is largely commoditized at this point. Lemonade can and does successfully stand out in terms of user interface, but cost leads represent a bigger opportunity to leap ahead of the pack. And thankfully, their AI-native, highly-malleable architecture simply lets them move faster and more efficiently.  This is already creating large productivity gains in marketing, underwriting, cross-selling, support and operations. It’s allowing LMND to 3x claim handling with no more people and no service issues and with ample LAE-based leverage. It’s allowing LMND to deliver the dramatic underwriting improvements unfolding. It’s allowing LMND to maintain that imperative 3x LTV/CAC while growing spend. And? It’s allowing LMND to routinely and rationally undercut the competition in a sector where customers are very sensitive to price… while greatly advancing its margin profile. That is the single most important way that LMND can sustainably differentiate, and it’s the byproduct of its AI-native plumbing. AI. Helps. Everywhere.

While LMND doesn’t get the credit that Palantir or others in the app layer of the AI opportunity get, perhaps it should. They’re using AI to create defensible cost advantages in a sector where that is make or break. And you don’t have to take my word for it. Just look at these financial trends.

Between LAE, loss ratio and other OpEx buckets, Lemonade feels like it’s in inning 1 of using AI to create a better and more profitable company. They think they’re miles ahead of other insurers, but with miles left of progress for them to enjoy as well.

How Will LMND Use This Cost Edge?

Lemonade is not planning on using all of these cost advantages to optimize for minimum near-term loss ratio. That’s not their aim. Their gross profit dollar north star is connected to lower loss ratios, but there are more factors at play. LMND can take advantage of superior efficiency by cutting prices on some plans in profitable ways that are hard to compete with. While that means higher loss ratios, customers also quickly react to lower pricing and make up for that with incremental demand powering profit growth. For this reason, and especially because loss ratios are now in such great shape, they may elect to pull pricing levers in coming quarters at the expense of GLR to accelerate gross profit dollar growth. I’m all for that. For now, they see loss ratios likely falling Q/Q again. They just wanted to alert investors about this because it’s probably going to happen sooner rather than later.

Car Insurance & Retention:

This product has officially taken the growth training wheels off. And while they were frustrated by that (in their minds) taking too long, it was the correct approach. Underwriting must be nailed down for a product like this to avoid ballooning losses. It’s gambling with horrendous odds to scale before that’s the case. Steady risk pricing improvements drove the strong loss ratio result and emboldened the team to meaningfully lean into more growth. Premiums for the segment rose 40% Y/Y for context. Acceleration is fully expected to continue as confidence grows and it expands to new markets. Around 50% of these plans continue to be from existing CAC-free customers, while multi-product adoption eclipsed 5% partially as a result of this traction. That is great for ADR, but aforementioned ongoing clean up of their home book (letting low quality plans expire) is masking that progress for now. That headwind will meaningfully fade middle of next year when comps are lapped. And with this added nuance, it’s even more impressive that they’ve ramped to 30% Y/Y growth.

Europe:

Customers in Europe rose 100% Y/Y as loss ratios fell materially. They expect to keep making underwriting improvements there faster than in the USA, as they have far more flexibility to adjust rates at their discretion. No filings or lengthy approval processes needed. The home business is quickly growing across several countries while growth in France is picking up and overall progress remains quite strong.

More Product Notes:

Very good quarter for Lemonade. This is the quarter when temporary headwinds finally dissipated enough to demonstrate just how strong structural trends are for the company. Accelerating growth…. quickly improving margins… dramatically better underwriting… comfortable levels of liquidity… significant momentum across every single product they offer. This wasn’t a fun name to own for a while. But? While the stock was broken, the company quietly kept executing.

They showed impressive control over their P&L and an ability to make and meet bold promises. This company has been guiding to the exact same profitability schedule for the last 3 years. Amid rampant inflation, soaring rates, hectic macro cycles and more… they just kept executing. This is the culmination of all that hard work, with so many reasons to believe things only get better from here. I’m not trimming any shares. What a fun day when a stock with a $16/share average cost goes up $20/share in a day.

Duolingo is a leader in language learning, with chess, music, math and literature offerings. While this is a learning-based platform, the learning is meant to be fun, competitive, social and engaging. This is a proven formula to keep users coming back. An obsessive determination to experiment and drive data-driven improvements to every single piece of its app is the foundation of its success. They’re always tweaking something to extract more engagement, monetization or something else. Constantly split-testing every single variable is in its DNA. It leads with product, rather than advertising and relies on word-of-mouth growth to power the vast majority of its success. It then supplements that with efficient marketing (from social media to the Super Bowl) to create viral moments and demand accelerants.

My Duolingo Deep Dive can be found here (the financials section is now dated).

User misses were actually a bit better than I was fearing, following all of the negative data we’ve covered over the last quarter.

On gross margin, outperformance was thanks to the AI cost headwind not being as large as expected. It still led to the Y/Y margin decline, but was more modest than analysts thought it would be thanks to cost deflation. They’re not focused on maximizing margin at this point. Cost optimization for all this new AI product work will come later with a proven playbook.

While annual bookings growth guidance was actually raised from 32% Y/Y to 33% Y/Y, the quality of that raise was quite low. It was thanks to currency tailwinds, as annual constant currency (CC) growth guidance for the year was maintained. Furthermore, most of the Q3 beat and currency help was offset by Q4 softness, which will understandably spook some people. Constant currency Q4 growth guidance for this vital forward-looking demand indicator was just 19% Y/Y. That’s a sharp and accelerating decline and is not good. While revenue is faring better with a small Q4 raise, that’s related to strong subscription bookings in recent quarters. Revenue growth staying strong relies on bookings growth staying strong, and Q4 is not strong. Much more on this throughout the next section.

Accelerating investments in AI and product innovation will likely lead to a slower pace of margin improvement next year. I think their commentary will drive down profit estimates a bit for 2026, as consensus currently expects pace of leverage to be the same as 2025. Duolingo trades for ~30x forward EPS based on the after-hours price action and expected revisions. EPS is expected to grow by 59% this year and by 27% in each of the next two years. I think there will be some negative guidance alterations for 2026 and maybe 2027 following this report. The 27% 2-year CAGR could fall to 26% or 25%.

User Growth Puts and Takes:

There are competing factors that led to Duolingo’s user growth again falling short of its guidance. On the bright side, their Luckin partnership led to exceptional growth in China while a batch of subtle product tweaks fostered rising platform retention overall. 

On the other hand, as I talked about in the earnings season preview and in recent coverage of this name, they have not fully rediscovered their U.S. social media mojo. As a reminder, they had to pull back from Duolingo’s “unhinged” content style that seemed to go viral every 5 seconds. Leadership mismanaged messaging about their AI-first pivot, and there was backlash. They’re still not back to their typical ways, but they have begun to lean back into edgier posts and are seeing “a lot of recovery.” While this may not seem important, it’s vital. Social media powers Duolingo’s growth engine and that engine has been out of gas for months. They’re finally re-filling the tank a bit, which will help future user growth. For now, this disruption is slowing them down.

The next item will be the most controversial part of this review. Duolingo is tweaking its strategic priorities for the next few quarters at least. It is moving modestly away from running its split testing engine to optimize for paid conversions, and will instead focus on faster DAU growth. This entails maximizing engagement rather than subscription bookings and means crafting product design in a way that skews a bit more towards “teaching better” vs. monetizing. They were quick to call this shift modest, but it is the source of the Q4 bookings reduction.

Why the Shift to DAU Prioritization?

People are going to run with this change as a sign that AI competition is displacing Duolingo. That’s not true. Duolingo has seen absolutely zero share loss gain from live translation products or from chatbot language lessons. Nada. Zilch. Their platform is far more engaging and actually has a full set of curriculum that can effectively teach to the point of someone being hired for a job in English. Beyond that, the sense of competition, social connection and entertainment Duolingo provides extends the lead vs. perceived competitors. Luis was quick to say that while “people like to tweet about it,” ChatGPT and Gemini are not impacting them in the slightest. Some people want to learn for fun… or for a job… or for a loved one… chat bots and live translators don’t change that reality.

Ok so why is it happening? I think there are two reasons. First, Duolingo is excited about all the new technology it can use to enrich its own product. It can use all of these assets to make sure it’s offering a more engaging, productive and broader platform. In doing so, it can maintain what makes its app better while still embracing this opportunity to improve even more.

Duolingo wants to teach as well as a human tutor and thinks expediting that process is more important than making sure they meet a quarterly bookings estimate. They believe emulating a more personalized human tutor can get them from 100M+ MAUs to billions as they round out their suite with many, many more subjects over time. Right now, making sure they’re laying the foundation to get to this product quality goal before anyone else is the top priority for making sure they grow DAUs at a healthy pace for decades, not just a few more years. With that said, they’re expediting this process at the near-term expense of bookings. It sounds like there were some concerns about DAU growth continuing to decelerate without making this pivot… so they’re acting.

The second reason is that Max adoption is disappointing them. Their current formula isn’t working as well as they’d like and they’re resetting. The hope is that they can spend a few months or quarters making Max more appealing while DAU growth accelerates. If done well, they can then turn that engagement boost into a subscription boost with a few quarter lag. More on some new tools they’re introducing to bolster this offering.

“Over the next few years, education and the way people learn are going to change fundamentally… We just see it in our own metrics and with how fast we can put out content with things like video call… Because of AI, we see we have line of sight now to create an app that can teach much better than anything that humanity has seen before.” – Co-founder/CEO Luis von Ahn

Duolingo began this strategic tweak in September and thinks the bookings impact will be a factor for at least part of 2026. They expect strong proof points to emerge in the meantime and think the change is why DAU growth stabilized in September and October.

“Financial impact from this kind of reprioritization is relatively small (the 3.4% bookings guidance reduction). We think that that's worth it because, as Luis said, it's a huge opportunity. So the risk/reward seems right. – CFO Matt Skaruppa

More on Max:

While hearing that Max adoption is below expectations is not good, it is worth noting that growth for the product was still over 100% Y/Y and this is now 9% of total subscriptions. And encouragingly, as initial cohorts come up for annual renewal, retention is slightly higher than Duolingo Super (the other tier). It’s doing well… Just not as well as expected.

There are two potential unlocks coming that should help this package accelerate. First, they’re testing the subscription and associated LLMs in China as we speak. They expect to get approval in the coming weeks and months, with plans to debut the package shortly after. This is already DUOL’s fastest-growing market and they’ve been delivering these results with a hand effectively tied behind their backs. This should build on promising retention and lifetime value trends in that important market as their ability to compete is unlocked.

The second unlock pertains to all Max Markets. Considering most subscribers are beginners, the popular Video FaceTime product is too advanced for most people to get value from. I speak a good amount of Spanish, but I can’t have fluent conversations. I need English cues to help guide me along the way. They just introduced guided video calls that blend the new language with a learner’s foreign tongue. That should make Max compelling for a lot more people, and they’re just about to start marketing the release.

Other Subjects:

Chess is thriving. It has millions of users and is already bigger than math and music. Retention is even better than its language learning product despite being around for a single quarter. That was impressive to hear. Head-to-head matches are off to a good start and will finish rolling out to remaining users in the coming weeks.

It doesn't sound like they’re as pleased with Math or Music. Math needs a lot more content, which is going to happen in a few months when Duolingo has the full K-12 common core curriculum. They're also “revamping” the music product.  Duolingo has no current plans to launch new subjects in 2026. There’s too much to do with existing categories.

More Notes:

The channel checks we’ve been working through over the last three months were spot on. DAU growth was again weak due to the same AI-first social media backlash that hurt them last quarter. And now… the small strategic pivot to prioritize engagement over monetization for the next few quarters adds more polarization to the investment case. I get it. They want to make sure the product suite is easily best positioned to take advantage of a massive, AI-led opportunity. They’re more interested in fortifying their product moat and extending their lead than optimizing for conversions and monetization to appease a short-sighted Wall Street. And I support that. This company has delivered too many amazing quarters and is too capable in the realm of product roadmap for me to give up on it because I’m annoyed with bookings guidance. Growth is still very good (22% bookings growth in Q4 is what they consider bad), multi-year operating leverage has been fantastic, its market share lead is still dominant, the category is humungous and the valuation is very reasonable. I’m not exiting.

But at the same time, I fully expect skeptics to only grow more emboldened following this report. They’ll assume the changes mean there’s something structurally off with DUOL’s business, and the Green Owl will need to show them otherwise.

I’m also not confident enough to add. This demand headwind is likely going to hold DUOL back a tad for most of 2026. And it creates a newfound layer of uncertainty for a company I thought was an execution machine. A highly capable leadership team makes me optimistic that they can fix their issues, but this report still created more investment risk. I now want them to prove it. I’m not willing to assume they will, following this negative surprise and the recent social media blunder they’re still recovering from. That’s two toe stubbings in short order, and pushes me to avoid buying this large dip.