Duolingo Finally Put a Number on the Pivot: $50 Million of Bookings for a Shot at 100 Million Users

Results vs. Consensus

Duolingo reported after the close on February 26. The quarter itself was a formality: on January 12 the company pre-announced Q4 DAU growth of approximately 30% and bookings at or slightly above the high end of guidance, and the February print confirmed both. What was not in the market was the 2026 guide, and that is where the entire reaction came from.

Year-over-Year Comparison

Full Year 2025

Sequential Comparison

Revenue

Revenue of $282.9M grew 35% year over year and beat the $276.8M consensus by 2.2%. On a constant-currency basis growth was approximately 33%. The full year finished at $1,037.6M, up 39%, and the four reported quarters sum exactly to that figure. The deceleration from 41% in Q3 to 35% in Q4 is the ratable-recognition lag beginning to work in the other direction, exactly as the Q3 bookings guide implied it would.

The forward path is where the number gets uncomfortable. Q1 2026 revenue is guided to $288.5M, up 25%, and the full year to $1,197M to $1,221M, up 15% to 18%. Management was explicit that the arithmetic of a 25% Q1 against a 15% to 18% full year means the back half runs below the low end of the annual range. That is a guide with a visible trough in it, disclosed rather than buried.

Margins

Gross margin of 72.8% expanded 90 basis points year over year, the first year-over-year gross-margin expansion in several quarters and a direct reversal of the AI-cost drag that framed the 2025 bear case. The full year came in at 72.2%, a decline of about 55 basis points against 2024, which is better than the roughly 80 basis points management guided to in November and much better than the 100 basis points guided in August.

That trend is being deliberately broken in 2026. Management guided gross margin to approximately 71% in Q1 and roughly 69% for the remainder of the year, a step down of about three points from the 2025 level, because AI features are being extended to a far larger share of the user base including free users. This is not cost inflation. It is a decision to spend gross margin on product access, and it is the single largest identifiable component of the 2026 Adjusted EBITDA guide.

Adjusted EBITDA of $84.3M on a 29.8% margin expanded 490 basis points year over year and was the second consecutive record. The full year finished at $305.9M on a 29.5% margin, up 380 basis points. Free cash flow of $93.7M grew 16%, though the margin compressed 550 basis points to 33.1% as the prior-year quarter benefited from an unusually strong collection profile. The full year generated $360.4M of free cash flow on a 34.7% margin.

Users, and the divergence that matters

DAUs of 52.7M grew 30% year over year against a 51% comparison, and rose 4.4% sequentially. MAUs of 133.1M grew only 14% and fell 1.6% sequentially, the first sequential MAU decline in the periods we track. The result is a daily-to-monthly engagement ratio of 39.6%, up 490 basis points year over year and 230 basis points sequentially.

That divergence is the most important unremarked fact in the print. A shrinking monthly base with a growing daily base means the users leaving are the marginal, low-frequency ones, while the core is deepening. Read charitably, it is evidence that friction drove off casual users and that the remaining cohort is more valuable per head. Read uncharitably, it is the top of the funnel narrowing, which is precisely the condition the 2026 plan exists to reverse. Both readings support the same conclusion about what the company should do; they differ only on how urgent it is.

Segment Performance

Duolingo reports as a single segment. The meaningful decomposition in Q4 is by subject, which management disclosed for the first time in a usable form, and by the user funnel.

Languages

The core business grew at roughly the pace of the total, which management stated directly when asked whether chess was masking language weakness. The 2026 plan invests here first: all courses teaching the nine most-learned languages are being extended to Duolingo Score 129, the level management equates to CEFR B2 and to employability in the language. Those nine courses cover more than 90% of daily active users, so the content investment reaches nearly the whole base rather than a niche.

"The way to see it is not that languages have not been growing. In fact, languages are growing at about the same pace as Duolingo because it's the largest course. It's just that we are getting some new users for chess, but a lot of them are just doubling up from languages and chess." — Luis von Ahn, Co-Founder and CEO

Assessment: The honest answer to a question designed to catch a cannibalization story. Languages are not decaying under the new subjects, and the advanced-content build arrives within a couple of months rather than as a multi-year promise. This is the least risky part of the 2026 plan and the part most likely to show up first.

Chess

Chess reached approximately 7 million daily active users in under a year, which management believes makes Duolingo the second largest chess property in the world. The detail that carries the most information is that the company has not yet done App Store optimization for chess, so the app does not surface in App Store chess searches at all. The 7 million came almost entirely from cross-selling the existing base.

"At this point, we believe we're the second largest chess thing in the world, which is, by the way, incredible because if you search for chess on the App Store, we don't appear in the results. We just haven't even gotten there to doing App Store optimization, but we already got to 7 million daily active users." — Luis von Ahn, Co-Founder and CEO

Assessment: This is the strongest single data point in the release. A subject built in nine months, distributed entirely through the company's own base, reaching second place in its category with zero acquisition effort, is a direct measurement of the value of the distribution asset. Management is now committing dedicated marketing to it, which converts a proof of concept into a growth channel.

Math

Management believes it will have the best K-12 math tutoring app this year and framed the addressable audience as roughly one billion learners, larger than any subject other than languages. It also named the honest problem: unlike language learners, math learners generally do not want to be learning math, and they sit in schools rather than on consumer app stores. The stated first go-to-market is consumer-supplemental rather than institutional.

"Our first step is to just make this product. We're almost there... At the moment, our hypothesis is that the easier way to get started. That may not be the ultimate thing we do, but the easier way to get started is to think of it as a supplemental thing for school. So rather than get deals with school districts and stuff like that, think of it more as a competitor to Kumon." — Luis von Ahn, Co-Founder and CEO

Assessment: The Kumon framing is a good one because it routes around the school-district sales cycle that has killed most education software. Management also flagged that parental willingness to pay for math is high, which makes this the one growth engine with a monetization story attached at birth. Nothing lands in 2026 numbers.

Advertising

Advertising is being repriced rather than expanded. Ad load will not increase, which management said would run counter to the entire 2026 strategy. Instead the company is moving from low-quality network inventory to direct deals, which raises both ad quality and revenue per impression. One experiment in test is serving part of an ad in the language the user is learning.

"What we're unlikely to do is increase ad load... However, we are doing a lot better of a job at displaying better ads... So I think that's kind of how it's going to be. We're experimenting with some stuff that is really cool, but I don't know if we'll make it work. I'll tell you, my dream has always been to be able to give you ads in the language that you're learning." — Luis von Ahn, Co-Founder and CEO

Assessment: Advertising is the one monetization vector that is fully aligned with the 2026 strategy, since it scales with the free base rather than taxing it. Moving from network fill to direct sales is a revenue-per-impression lever that does not consume any of the friction budget. It is small today and under-modelled.

Key Topics & Management Commentary

Overall Management Tone: The call was the most quantified and least evasive of the past several quarters, which is notable given that it delivered the worst guide. Management pre-empted rather than defended, opening with the strategy, naming the dollar cost of it, offering explicit modelling guidance including a warning that back-half growth runs below the annual range, and stating a new philosophy of narrowing the gap between guidance and actuals. The one place the answers thinned was on when any of this shows up in the data, where the honest response was that it will not, this year.

1. The 2026 guide and what it concedes

The guide is the news. Bookings are guided to grow 10% to 12% against 33% in 2025 and 24% in Q4 alone. Revenue is guided to 15% to 18%. Adjusted EBITDA of $299M to $305M is flat to slightly down against 2025's $305.9M, which means the company is guiding to no profit growth at all in a year in which revenue grows more than 15%.

"At the highest level, our 2026 guidance is bookings growth of 10% to 12%, revenue growth of 15% to 18% and adjusted EBITDA margin around 25%." — Gillian Munson, Chief Financial Officer

The CFO also volunteered the shape rather than leaving it to be discovered: Q1 bookings were tracking above guidance quarter to date as of the prior Friday, the first half faces the hardest comparisons of the year, and the revenue growth rate steps down through Q2 and Q3 before stabilizing.

Assessment: A guide this far below the Street, delivered with this much detail and this little hedging, is a different artifact from a guide-down that gets extracted in Q&A. The information content is high and the surprise risk for the rest of the year is correspondingly lower.

2. The $50 million number

In November, management described the cost of the strategy shift as "not a humongous change," "relatively small," and "not that big," and declined three separate invitations to size it. That was the specific reason we initiated at Hold. This quarter the number arrived, in the shareholder letter, unprompted.

"To quantify this tradeoff, we estimate we're investing more than $50M of foregone bookings from friction (or about 5 points of year-over-year bookings growth) into the free user experience to drive word of mouth and user growth." — Luis von Ahn, Co-Founder and CEO

The CFO then decomposed the remaining gap when pressed on it, attributing roughly half of the deceleration from Q4's 24% bookings growth to the guide to continued user deceleration and the other half to the deliberate monetization pullback.

Assessment: This is the disclosure that changes the investment case. A five-point voluntary haircut to bookings growth is a number a shareholder can accept, reject, or model a reversal of. An unquantified "years" of investment is not. The company converted an open-ended risk into a bounded, dated one.

3. Video Call moves to Super

The most concrete product decision of 2026 is moving Video Call, the flagship feature of the top-priced Max tier, into the cheaper Super tier, which has roughly ten times as many subscribers. Management framed this as an offensive move enabled by falling inference costs, explicitly denied that Max performance prompted it, and acknowledged it does not know what the A/B test will do to bookings.

"It may be that this increases revenue because more people sign up to Super Duolingo or because the retention of Super Duolingo is significantly higher because video call is there, or it may be that it decreases bookings because we're not making as much money from selling Max. I don't know what it will do." — Luis von Ahn, Co-Founder and CEO

Management pre-committed to a fallback, metering Video Call in Super to a fixed number of sessions per day if the Max cannibalization is severe, and said the outcome will stay within the bounds embedded in the guide.

Assessment: Giving away the premium tier's differentiating feature is the boldest decision in the plan and the one with the widest outcome distribution. The pre-committed metering fallback is what makes it underwritable: management has told you the downside is bounded and named the mechanism. The upside case, that conversational practice becomes a retention feature for ten times as many subscribers, is not in anyone's model.

4. Gross margin is being spent on purpose

Gross margin is guided to approximately 71% in Q1 and roughly 69% thereafter, against 72.2% for 2025. That roughly three-point step down is entirely attributable to extending AI features, chiefly speaking practice, to free users. Management is introducing voice answers for most exercises across free and paid tiers, plus a new free lesson type called Speaking Adventures, scaling around midyear.

"As you think about our COGS, the key thing to keep in mind is that we plan to share AI features with a far greater portion of our user base. This is anticipated in the lower gross margins that we outlined in our letter." — Gillian Munson, Chief Financial Officer

Assessment: Three points of gross margin on roughly $1.2B of revenue is on the order of $36M, which is a large fraction of the total 2026 profit sacrifice. It is also the most reversible piece: inference costs have fallen every quarter for two years, and a cost curve that keeps falling gives this margin back without any product decision being unwound.

5. The monthly-user decline nobody asked about

MAUs fell from 135.3M to 133.1M sequentially and missed the 138.0M consensus by 3.5%, while DAUs rose. The word "MAU" does not appear in the call transcript. The only reference to monthly users was management's observation that about 10% of them pay, made in the context of the monetization runway rather than the metric's direction.

The mechanical read is consistent with everything else management said: friction drove casual monthly users away faster than it converted them, leaving a smaller, denser base. The company's own diagnosis of the 2025 problem predicts exactly this outcome.

Assessment: Eleven analysts asked questions and none raised a metric that declined sequentially and missed by 3.5%. If the 2026 plan works, MAU turns before DAU growth reaccelerates, because it is the top of the funnel. It is therefore the single best leading indicator available for grading this strategy, and it is currently unwatched.

6. The 100 million DAU goal and the two-path math

Management set a medium-term target of 100 million daily active users in 2028, roughly double the current base, and the CFO put arithmetic behind both branches of the decision in a way that is unusually explicit for a public company.

"We could have grown the business, probably we would be at about a mid-teens CAGR over the next couple of years. If we could keep our margins where they are, we're probably looking at a $1.5 billion business, $400-plus million in adjusted EBITDA. But we, at the company, are really motivated to go for the bigger prize... if we can get to that DAU and even reasonable monetization assumptions from where we are today, and we believe we can scale our expenses over time, you're looking at a business in a couple of years, it could be $2.5 billion with over $700 million in adjusted EBITDA." — Gillian Munson, Chief Financial Officer

The CFO labelled this "back of the envelope," which it is. It is nonetheless the first time management has framed the trade as a quantified fork rather than as a philosophy.

Assessment: The comparison is doing a lot of work and should be treated as an argument rather than a forecast. What matters is the disclosure of the counterfactual: management is telling shareholders what the safe path was worth, which is a level of candour that makes the risky path assessable rather than merely asserted.

7. The $400 million buyback

The Board authorized a repurchase of up to $400 million, announced in the same release as the guide. Against the post-reaction market value that is roughly 8% of the equity. The CFO said the company expects to execute on the authorization during the coming year, which is stronger language than a typical authorization carries.

"I would like to highlight our buyback authorization that was announced today. Our Board has authorized a buyback of up to $400 million in our shares. We believe this represents good capital allocation discipline and expect to execute upon this authorization in the coming year." — Gillian Munson, Chief Financial Officer

The context is that 2026 stock compensation is guided at almost 15% of revenue against 13% in 2025, with fully diluted share count expected to rise 3.5% to 4% before any buyback effect. The authorization is therefore partly a dilution offset and partly a genuine return.

Assessment: The absence of any capital-allocation framework was on our list of omissions from the November call. It has been answered within one quarter, with a size that is material against a de-rated equity. Not a single analyst asked about it.

8. The CFO transition and a new guidance philosophy

Gillian Munson, previously a Board member and Chair of the Audit, Risk and Compliance Committee since 2019 and formerly CFO of Vimeo, took over from Matt Skaruppa effective February 23, three and a half days before this call. Skaruppa remains in an advisory role through November 2026. The company stated that neither departure involved any disagreement over operations, policies or practices.

The new CFO used her first call to state a change in guidance philosophy directly.

"You may ask about our guidance philosophy. And so I thought I'd note that. This year, our view is that like in Q4, our hope is to narrow the gap between our guidance and actuals versus what you have generally seen from Duolingo in the past." — Gillian Munson, Chief Financial Officer

Assessment: This materially changes how the 2026 guide should be read. Duolingo has beaten the top end of its own ranges in each of the last three quarters, and the market has learned to add a cushion. A stated intent to guide closer to the line means less cushion should be added, which makes an already-low guide read lower still. It also means a miss would carry more information than it used to.

9. Saturation and competition, tested against data

Asked whether market saturation or competitive intensity explained the 2025 deceleration, management answered with penetration statistics rather than assertion: 2% of daily internet users in the United States use Duolingo, 3% in the United Kingdom and 4% in Germany, with Germany not being the highest. Bringing every country to the US level alone would more than double the daily base. On competition, Duolingo holds roughly 85% of global language-app daily active users, a share management said has been flat for some time.

"Even if we only assume that every country got to 2%, which is the U.S. penetration, we would more than double our daily active users. So I just don't think we're near saturation at all... For example, we survey our churn users. And the answers to the surveys of our churn users haven't really changed in years." — Luis von Ahn, Co-Founder and CEO

Assessment: The strongest empirical answer given on the call, and it is falsifiable. If churn-survey composition has not shifted in years, then AI substitution is not showing up where it would first appear. The penetration math independently supports management's claim that 100 million daily users is a distribution problem rather than a demand ceiling.

10. Why a new entrant has not appeared

On whether AI lowers the barrier for a competitor to be built quickly, management made two arguments: that the same tooling accelerates Duolingo, and that the binding constraint is distribution rather than construction. The specific mechanism offered is that a new entrant cannot make performance marketing pay against a free product as good as Duolingo's, so it paywalls, which caps it.

"What they usually do is they try performance marketing to try to grow fast, but because our free product is so good, they usually cannot make their performance marketing work if they are to have a free product. So what then they do is they put a paywall on it. And then that just really caps them." — Luis von Ahn, Co-Founder and CEO

Assessment: This is the cleanest statement of the moat we have heard from this management team, and it is internally consistent with the 2026 strategy: if the free product's quality is what makes competitor acquisition economics fail, then spending $50M to improve the free product is a defensive investment as much as an offensive one. That reframing is not in the market's current reading of the guide.

11. Over-monetization, admitted

Management conceded the diagnosis directly: the company is simultaneously undermonetized in penetration terms, with only about 10% of monthly users paying against roughly half at a comparable freemium consumer subscription business, and overmonetized in mechanism terms, because the fastest lever it had found was adding friction.

"We got ourselves into an interesting situation where both Duolingo is undermonetized and overmonetized at the same time... the way we were increasing monetization was we found that the quickest way to increase monetization was basically by adding friction. The more we added friction, the more we got people to subscribe. And that's okay, but I think that we got it to a point where it really became at odds with DAU growth." — Luis von Ahn, Co-Founder and CEO

The replacement mechanisms named are avatar customizations, a native ad format and direct ad sales, all of which scale with the base rather than taxing it.

Assessment: A management team that publicly diagnoses its own error and names the replacement levers is more investable than one that does not, provided the levers are real. Avatar customization and native ad formats are proven mechanics in adjacent categories. The honest caveat management gave is the right one: these take longer to build than doubling the ad load.

Guidance & Outlook

The guide is unambiguous about what it is: a decision to hold profit dollars flat for a year while spending the growth. Adjusted EBITDA of $299M to $305M against 2025's $305.9M means every incremental dollar of the roughly $170M of revenue growth is being consumed. The 450 basis point margin compression comes from three identifiable places: roughly three points of gross margin from extending AI features to free users, faster-than-revenue growth in R&D and sales and marketing, and the more than $50M of foregone bookings from friction removal, which never enters the revenue line at all.

Implied Q-over-Q ramp: Q1 bookings of $301.5M against Q4's $336.8M is a normal seasonal step down from the New Year's promotional quarter. The full-year midpoint of $1,286M less Q1's $301.5M leaves $984.5M across the final three quarters, an average of $328M per quarter. Management guided the Q2 bookings growth rate to fall slightly more than three points from Q1's 11%, then rise in the second half, so the profile is trough-in-Q2 with a modest recovery embedded late. On revenue, the company stated outright that Q3 and Q4 growth run below the 15% low end of the annual range, given a 25% Q1.

Street at: Consensus sat near $1.264B for 2026 revenue against a guide midpoint of $1.209B, roughly 4.4% below. On bookings the gap is wider in growth terms: the Street was extrapolating from a 24% Q4 and received 10% to 12%. On Adjusted EBITDA there was no reliable two-source pre-print figure, but a Street modelling continued margin expansion off 29.5% would have been carrying something well above $305M.

Guidance style: Historically conservative, and now explicitly less so. Duolingo cleared the top end of all six guided ranges this quarter and has done so repeatedly. The new CFO's stated intention to narrow the gap between guidance and actuals means the customary cushion should be reduced. Set against that, she also disclosed that quarter-to-date Q1 bookings were tracking above the guide as of the prior Friday, which is a cushion signal in the opposite direction on the very first quarter of the new regime.

Analyst Q&A Highlights

Why now, and what actually gets to 100 million daily users

The opening question asked for the mechanism rather than the philosophy: what specifically about the AI landscape makes this the moment, and which one or two product priorities carry the doubling. The answer named three investment buckets and, more usefully, produced the chess disclosure that reframes the whole distribution argument.

Q: "Can you just elaborate a bit more on what you're seeing in the market across AI and just overall innovation across the technology space that makes this the right time for the strategic shift to reaccelerate user growth? And I guess, more fundamentally, to achieve your 2028 target to double users, what are the, call it, 1 to 2 top priorities across the product that will really enhance that DAU growth and drive the implied acceleration to get there?"
— Bryan Smilek, JPMorgan

A: "In general, the 3 buckets that we're going to be investing in are teaching languages better... Another one is in making the free user experience better. The less friction that there is in the user -- in the free user experience, the more users we have. We have seen this over the last 15 years. Whenever we'd remove friction from the free user experience, DAUs grow, right?"
— Luis von Ahn, Co-Founder and CEO

Assessment: The claim that removing friction reliably grows DAUs is presented as a fifteen-year empirical regularity, not a hypothesis. If that is true, the 2026 plan is closer to running a known playbook in reverse than to a speculative bet, which is a materially lower-risk framing than the market is applying.

Whether something changed since the last call

The sharpest question of the call noted that November's framing implied a smaller 2026 impact than what was actually guided, and asked what had changed in the interval. Management identified two things without deflecting.

Q: "Last call, when you talked about the reprioritization for 2026, and I think it was sort of contemplated it was going to have some impact on the financials for 2026. But maybe not to the level guided that was presented today... did something change since last quarter in terms of how you thought about 2026?"
— Ralph Schackart, William Blair

A: "What happened between then and now are two things. The first is that we kept on seeing this trend of user deceleration. And by now, we expect about 20% year-over-year growth throughout 2026. And that -- and the other thing is that my conviction of AI just increased even more."
— Luis von Ahn, Co-Founder and CEO

Assessment: One of the two reasons is a data point that got worse and one is a conviction that got stronger. That is a fair answer, and it also concedes that November's characterization of the impact as "not that big" was wrong. The deceleration is doing more of the work than the strategy is, which is the less comfortable half of the explanation.

Decomposing the bookings guide

A question on how the guide treats the Super and Max experiments produced the most model-relevant answer of the call: an explicit split of the deceleration between what is happening to the company and what the company is doing to itself.

Q: "On the bookings outlook, just given some of the unknowns you spoke to... on how video call tests are going to play out on Super. What is the bookings outlook factor in terms of conversion rate of Super relative to what you've experienced in the past?"
— Alexander Sklar, Raymond James

A: "If you take -- start from Q4 bookings growth of about 24% and then look over to the guide, we think about half of that is the user deceleration continuation that Luis talked about. And then the other half of that is us monetizing more carefully and thinking about taking away some friction for our users."
— Gillian Munson, Chief Financial Officer

Assessment: A clean fifty-fifty split of a thirteen-point deceleration. Half is discretionary and reversible at management's option; half is not. That decomposition is what makes the 2026 guide analyzable rather than merely disappointing, and it arrived from a CFO on her fourth day.

Saturation and competition as the alternative explanation

A line of questioning tested the two explanations management had not offered for the deceleration, and got penetration data rather than reassurance.

Q: "You mentioned these factors that caused this deceleration, law of large numbers, increased maybe over focus on monetization. There are 2 other potential reasons, too. So just address those, which is market saturation and competitive intensity."
— Mark Mahaney, Evercore ISI

A: "In terms of competition, if you look at language learning apps, we have about 85% of the daily active users of language learning apps in the world. And that number has remained pretty flat over a while... in the United States, 2% of all Internet users on a given day use Duolingo. So 2% penetration in the U.S. for daily users. In the U.K., that's 3%. In Germany, that's 4%."
— Luis von Ahn, Co-Founder and CEO

Assessment: The response is data rather than narrative, and it is the kind of data that would look different if the bear case were right. A flat 85% category share is inconsistent with competitive erosion; a 2% domestic penetration rate is inconsistent with saturation. The remaining explanation is the one management gave, which is that it taxed its own funnel.

What is embedded in the guide for the Super and Max change

A follow-on pressed on how the guidance treats an experiment whose outcome management admits it cannot predict. The answer was the concept of guided bounds plus a named fallback mechanic.

Q: "As you think about the guidance for fiscal '26, I guess what assumptions, if any, are you making sort of building in for when video call rolls out to Super for sort of... ARPU trend changes of maybe downsells from Max down to Super from that?"
— Ryan MacDonald, Needham & Company

A: "We put in our guidance some bounds, what we think that will actually -- this is what we think in all, and there are some bounds. We'll stay within those bounds. And we have different ways of staying within those bounds. So for example, if we see that putting video call in Super really destroys our Max business or something, which is like really terrible, then we'll start doing something like metering video call in Super."
— Luis von Ahn, Co-Founder and CEO

Assessment: The answer nobody wants and the right one. Management is not forecasting the A/B test; it is committing to a guardrail and naming the tool. That converts an unbounded product risk into a bounded one, and it is why the Super change should not be modelled as an incremental hit on top of the guide.

What monetization looks like on the other side

One line of questioning skipped past 2026 entirely and asked what the company will have learned, and how monetization resumes once the free experience has been rebuilt. The answer was an unusually direct admission of the prior error.

Q: "It sounds like one of the lessons was that you guys essentially overmonetized your user base. So Luis, as we think about going through this transition, what happens on the other side in terms of the lesson that you guys now have as you guys come back and then start to monetize users later on?"
— Andrew Boone, Citizens JMP

A: "Only about 10% of our monthly active users pay us. We wholeheartedly believe we can do much better than that... These other ways of monetizing are just going to take a little longer to do, but we're very confident that we're going to have some because only 10% of our active users pay us to subscribe."
— Luis von Ahn, Co-Founder and CEO

Assessment: The exchange establishes that the monetization ceiling is not the constraint; the mechanism was. A 10% paid penetration against a stated comparison of roughly half at a peer freemium subscription business is the size of the prize, and it is the reason the friction sacrifice is affordable rather than terminal.

Whether AI lets a new competitor appear

The final substantive exchange raised the version of the AI risk that is hardest to dismiss: not that an existing rival wins, but that the cost of building a rival collapses. Management answered on distribution economics rather than on product.

Q: "What AI could potentially do is accelerate the path to creating new competitors that don't even exist today. So could you talk to potential risk of a competitor emerging because AI is making it easier to create certain products?"
— Shweta Khajuria, Wolfe Research

A: "The reality is that if I were to write down just the spec of Duolingo, just -- and I'm not talking about recreating it in a programming language, like write it down in English, it would take me years because there are 1 million corner cases of things that we do that we're really smart about... over the span of Duolingo, we have probably seen, I don't know, 400 apps come up and try to compete with us. And I'm making that number up, its hundreds. And we just don't see them grow again because we have a very large distribution."
— Luis von Ahn, Co-Founder and CEO

Assessment: The strongest formulation of the moat argument this management team has given, and the one that ties the 2026 spending to defense rather than only to offense. The specification argument is soft; the distribution argument, backed by a flat 85% category share across hundreds of entrants, is not.

What They're NOT Saying

1. The sequential monthly-user decline: MAUs fell from 135.3M to 133.1M and missed consensus by 3.5%. The metric was never named on the call by management or by any of the eleven analysts who asked questions. It is the leading indicator for whether the 2026 plan is working, and it is currently the least-discussed number in the release.
2. Any 2027 or 2028 financial framing: the 100 million DAU goal is dated 2028, but the only financial content attached to it is a CFO scenario explicitly labelled back-of-the-envelope. There is no committed margin trajectory, no monetization-per-user assumption and no capital-intensity framing for the target year.
3. What "modest early returns" actually means: management said the guide embeds only small late-year benefit from the investments and that shareholders should not expect visible acceleration until then. It never quantified how much benefit is embedded, so a beat and a miss on that component are indistinguishable.
4. Max subscriber count and economics: Q3 disclosed Max at 9% of subscribers with bookings doubling. This quarter, on the call where the tier's flagship feature is being given away, no updated Max share, renewal rate or ARPU contribution was provided. Management asserted "the Max numbers are actually quite good" without a figure.
5. The Duolingo English Test: not mentioned once on the call, for the second consecutive quarter, and no revenue disclosure was given. The Duolingo Score credentialing ambition described at length last quarter went unmentioned.
6. China: described in November as the fastest-growing country, second largest by DAUs, roughly 5% to 6% of the business, with Max in regulatory testing and launch expected in weeks to months. Not mentioned once this quarter, and no update on whether that launch happened.
7. Family plan: disclosed in November at roughly 29% of subscribers. No update, and no mention of how the New Year's promotion performed against the prior year's unusually strong family-plan cohort.
8. Taxes, for a second consecutive quarter: the words "tax" and "valuation allowance" appear nowhere in the call, despite a $256.7M full-year benefit and a guided 2026 effective tax benefit of approximately 6% that appears only in the letter. Nobody asked what the normalized cash tax rate looks like once the shield runs out.
9. Buyback execution parameters: a $400M authorization was announced and no analyst asked a question about it. No pace, no price discipline, no statement of whether it is intended to fully offset the guided 3.5% to 4% dilution or to reduce share count outright.
10. The cost of the CFO transition: the new CFO joined three and a half days before delivering a guide that reset the equity. Management did not address whether the 2026 plan predates her arrival, and no analyst asked whether a departing CFO and a reset guide in the same quarter were related. The filings state neither resignation involved any disagreement over operations, policies or practices.

Market Reaction

• Pre-print setup: DUOL closed at $117.45 on February 26, down 33.1% year to date against the S&P 500's +0.9%, down 69.6% over the trailing twelve months and down 19.6% over the trailing thirty days. The 52-week closing range entering the print was $106.14 to $540.68.
• Next-day session: the stock opened at $92.26 on February 27, 21.4% below the prior close, traded a range of $91.99 to $101.45, and closed at $101.00, down 14.0% or $16.45. The close was a new 52-week closing low.
• Recovery within the session: the stock closed 9.8% above its intraday low and near the top of its daily range, reversing more than a third of the opening gap.
• Volume: 20.1M shares against a 30-day average of 2.5M, or 8.1x normal.
• Market context: the S&P 500 fell 0.4% on the session, so the move was almost entirely idiosyncratic.
• Sell-side action: a cluster of downgrades and steep target reductions landed the same morning, framed around the 2026 guide rather than the Q4 print.

The intraday pattern is the informative part, and it is the opposite of November's. On the Q3 reaction day the stock gapped down 25.3% and closed at 25.5%, meaning it finished at the bottom of its range with no bid. This time the stock gapped down 21.4%, traded as low as 21.7% down, and then recovered to close down 14.0%, near the session high. Roughly a third of the opening gap was bought back during regular hours.

That is a different kind of session. A gap that holds all day is repricing; a gap that gets bought back is forced selling meeting a buyer. At 8.1x normal volume, both sides were sized. The most plausible reading is that the overnight move was mechanical, driven by growth mandates that cannot hold a business guiding to 11% bookings growth, and that value-oriented capital met it at a level where net cash is 23% of the market value.

It is also worth being precise about what the print did and did not contain. The Q4 numbers were pre-announced on January 12 and confirmed. The strategy was announced on November 5. What was genuinely new on February 26 was the magnitude: 10% to 12% bookings growth and flat Adjusted EBITDA dollars. The market had accepted the direction three months ago and repriced only when the arithmetic arrived.

Street Perspective

Debate: Is this the bottom of the reset or the first of several?

Bull view: The bull case being made is that everything is now on the table. The strategy was disclosed in November, the Q4 numbers were pre-announced in January, and the cost was quantified in February at more than $50M and about five points of bookings growth. A new CFO has reset the guidance philosophy on her first call, which is the single best moment in a company's life to clear the deck. There is nothing obvious left to disclose.

Bear view: The bear camp contends that the November guide was supposed to be the reset and was not, that management itself admitted its prior characterization understated the impact, and that half the deceleration is user-driven rather than discretionary. If 2026 DAU growth lands below 20%, the 2027 guide resets again from a lower base.

Our take: The bulls have the better position, mostly on the strength of the CFO change. The one structural reason a guide gets reset twice is that the first reset was managed to protect a number; the person who set the November guide is now an advisor, and the person who set this one used her first call to say she intends to guide closer to actuals. That does not eliminate the risk of a 2027 reset, but it removes the mechanism that produced this one.

Debate: Is the pivot a strategy or a rationalization of decay?

Bull view: The optimistic reading points to the evidence management put on the table: 85% of global language-app daily users, flat over time; 2% penetration of US daily internet users against 4% in Germany; churn-survey composition unchanged for years; and chess reaching 7 million daily users in under a year with zero App Store optimization. None of that looks like a business losing to substitution.

Bear view: The skeptics point out that DAU growth has gone 40%, 36%, 30%, and is now guided to 20%, that MAUs are falling sequentially, and that a company genuinely confident in its funnel does not need to give back $50M of bookings to defend it. The strategy and the decay are observationally equivalent from outside.

Our take: The MAU decline is the bears' best evidence and the DAU/MAU ratio is the bulls'. Both are true simultaneously, and the combination favors the bull reading: a base shrinking at the monthly margin while deepening at the daily margin is the signature of a company that taxed its casual users, not one whose product is being substituted. Substitution would show up as daily users leaving, and daily users grew 4.4% sequentially.

Debate: Does the valuation do the work?

Bull view: At $101.00 the enterprise value is roughly $3.9B against 2026 guided bookings of $1,286M and guided free cash flow above $350M. That is about 3.0x forward bookings and about 11x forward free cash flow for a business with no debt, 23% of its market value in net cash, an authorized buyback worth 8% of the equity, and management guiding 20% daily-user growth in what it calls its worst financial year.

Bear view: The bear response is that free cash flow is flattered by deferred revenue growth that decelerates with bookings, that stock compensation rising to almost 15% of revenue makes the Adjusted EBITDA number a poor proxy for owner earnings, and that a 3.5% to 4% increase in diluted shares absorbs a large part of what the buyback returns.

Our take: The dilution point is the serious one and it is real: on the guided numbers, the buyback is closer to a dilution offset than a return of capital. But the bear case still requires the multiple to compress from 3x forward bookings, which is a level that already embeds permanent low-teens growth. The market is pricing the trough year as the terminal state, and management has told you which half of the deceleration it controls.

Model Update Needed

Valuation impact: At the $101.00 reaction close and 49.8M estimated diluted shares, the equity is worth roughly $5.0B. Netting $1.14B of cash and short-term investments gives an enterprise value near $3.9B, or approximately 3.0x the FY26 bookings guide midpoint, 3.2x the FY26 revenue guide midpoint and 12.9x the FY26 Adjusted EBITDA guide midpoint. Against guided free cash flow of more than $350M the multiple is about 11x. Our prior target was $215 against a materially different bookings trajectory; we cut it to $140, which applies roughly 4.5x to the FY26 bookings guide midpoint and 19.3x to guided Adjusted EBITDA, and implies +39% from the reaction close.

Thesis Scorecard Post-Earnings

Overall: The thesis is unchanged in substance and materially clearer in shape. Two of the three bull pillars were confirmed outright and the third was suspended by choice rather than damaged. Of the two bear points that drove our Hold, one escalated and one was largely answered. The single change that matters is that Bear #2, the unquantified open-ended investment, is no longer unquantified, which was the specific condition we said in November would need to resolve before we could underwrite the multiple.

Action: Upgrade to Outperform from Hold. We said in November that we would rather pay up after the February guide than guess before it. The February guide came in worse than we expected and the price came down more than the guide did: bookings growth is roughly half what we had modelled, but the enterprise value is 54% lower than at our initiation and the stock now trades at 3.0x guided forward bookings with 23% of its market value in net cash and a buyback authorized at 8% of the equity. Management has quantified the sacrifice, dated the payoff, decomposed the guide, and put a new CFO on record intending to guide closer to actuals. The remaining risk is execution against a 2028 goal, which is a risk worth taking at about 11x free cash flow and was not one worth taking at the multiple the stock carried in November.