Duolingo (DUOL) Q3 2026 earnings review
Duolingo's user growth speeds up, but each user pays less
Duolingo's focus on free users is bringing in far more people, and bookings have begun to follow, though paying subscribers lag. Daily users grew 29% from a year ago, the second acceleration in a row, and bookings growth climbed back to 12% from 8% in the spring. Chief executive Luis von Ahn said the company will now gradually put more focus on bookings, and finance chief Gillian Munson raised every full-year target.
| Daily active users | 64.9M +29% from a year ago; +23% last quarter |
|---|---|
| Bookings | $316.1M +12% from a year ago; +8% last quarter |
| Paid subscribers | 13.1M +14% from a year ago |
| Full-year bookings outlook | Raised $11M $1,296M, about 12% growth |
⚖️ Verdict: 🟢 Bullish
The story got better because the user growth Duolingo paid for all year is now turning into bookings, and the company lifted every full-year target. The bad news: each daily user brings in about 13% less money than a year ago, and paid subscribers grew only 14%, so the paying base lags far behind the audience. Max, the top tier, is gone, and the claim that this costs little rests on numbers Duolingo does not publish.
The question now is whether the new users will start paying at the old rate or whether Duolingo must settle for a bigger but poorer audience. One answer says conversion takes time, because people subscribe months after they arrive; the other says longer free trials and game tie-ins pull in people who were never going to pay. Fourth-quarter bookings and paid subscribers, due in February, will tell.
🐂 Bull Case
User Growth Speeds Up for a Second Quarter
Duolingo chose this year to grow its free audience first and earn from it later. Chief executive Luis von Ahn repeated the logic this quarter: "Every new learner expands the audience we can monetize over time". Daily active users, people who open the app on a given day, are the measure the company manages to.
Daily users grew 29% from a year ago, up from 23% in the spring quarter. Von Ahn credited hundreds of experiments on notifications and streaks, plus tie-ins with two mobile games, Brawl Stars and Eggy Party, which "brought us millions of new and returning users".
- Daily users: 64.9 million, up 6.2 million since June by our math, against a 2.8 million rise a year earlier
- Daily users as a share of monthly users: 42%, from 37% a year ago, by our math
The rising share means existing users return more often, the kind of growth a campaign cannot buy. But the language app again did not size what the game tie-ins brought, just as it never sized June's streak event, so we cannot separate lasting gains from the campaign effect.
What to watch: fourth-quarter daily users in February, which von Ahn put "similar to Q3". Growth near 29% again would show the experiments carry the gains on their own; a slide back toward 23% would mean the game tie-ins did much of the work.
Bookings Growth Rises Again After Four Quarters Falling
Bookings, the cash customers pay up front for mostly year-long subscriptions, lead revenue by a few quarters. Their growth had slowed for four straight quarters, to 8% in the spring, while reported revenue still grew on subscriptions sold earlier.
This quarter bookings grew 12%, led by subscription bookings up 16%, which finance chief Gillian Munson tied "in large part to the strength in user growth". The result came in 3% above Duolingo's own target, the largest beat of the year by our math, and the fourth-quarter plan calls for 13% growth.
Revenue growth slowed to 14% from 18%, as Munson said it would, because the app spreads subscription revenue over a year. "As bookings growth reaccelerates, we expect revenue growth to follow," she wrote, and her fourth-quarter plan has the two rates nearly meeting.
What to watch: fourth-quarter bookings in February, against a year-ago quarter that itself grew 24%. Landing on the plan would confirm the slowdown ended in the spring; a shortfall would put that back in doubt.
Cheaper AI Pushes Gross Margin Above Plan
Duolingo's main cost of serving a lesson is computing: the AI models and hosting behind features such as Video Call, its AI conversation practice. Von Ahn said last quarter that a call now costs under a cent, down from about 30 cents, after a move to open-source models.
Gross margin beat the language app's own plan, which Munson credited to "lower feature-based AI costs and favorable hosting cost trends". The company is spending the savings: Video Call goes to all Super subscribers by year-end, and the fourth-quarter plan has margin slipping "as we expand the rollout of AI-based features on our platform". The risk stands, because other firms set the model prices Duolingo pays.
- Gross margin: 72.9%, against 71% planned and 72.5% a year ago
- Cost of revenue: 27.1% of revenue, from 27.5%, by our math
- Fourth-quarter plan: 72.0%, against the 69% set in February
What to watch: fourth-quarter gross margin against the 72.0% plan. Holding it with Video Call open to every Super subscriber would show the savings outrun the extra use; a miss would mean free features cost more than management thinks.
🐻 Bear Case
Each Daily User Brings In 13% Less Money
Duolingo's plan rests on free users turning into paying ones later. For now the ratios run the other way, because the audience is growing far faster than the money it produces, and the gap widened again this quarter.
- Bookings per daily user: $4.87, down 13% from a year ago by our math, after a 12% drop last quarter
- Paid subscribers: 13.1 million, up 14%, after 17% growth last quarter
- New paid subscribers in the quarter: 0.4 million, against 0.6 million a year earlier
Part of this is by design. Longer free trials and the Video Call giveaway hold back paid conversions while lifting daily use, and the company said in February it would favour users over money this year. But the paid base added fewer people than a year ago while the user base added far more, so the distance between the two keeps widening.
Von Ahn said Duolingo is now "gradually putting more focus on growing bookings in ways that preserve the quality of our free experience". Subscription bookings did accelerate, so existing payers are spending more or renewing in greater numbers, which is the one sign in the money lines that conversion is on its way.
What to watch: paid subscribers at year-end, now that Max subscribers move into Super family plans. Subscriber growth climbing back toward user growth would mean conversion only lagged; growth stuck near 14% would mean the app is building an audience it cannot charge.
Everything Outside Subscriptions Is Shrinking
Duolingo earns money four ways: subscriptions, ads shown to free users, the Duolingo English Test, a paid online English exam, and in-app purchases of virtual goods. Subscriptions carried the quarter and now make up 87% of revenue by our math; the other three lines together went backwards.
- Non-subscription bookings: $37.7 million, down 9% from a year ago, by our math
- Advertising revenue: $20.8 million, down 1%
- In-app purchases: $6.9 million, down 38%
- Duolingo English Test: $9.9 million, up 2%
Munson nonetheless based part of the higher fourth-quarter outlook on "continued strength in advertising and in-app purchases", which the revenue table does not show. The likely cause sits in the strategy itself: a user on a free trial sees no ads, von Ahn said last quarter, so the fastest-growing part of the audience earns nothing from advertising.
His answer for purchases is collection events such as Junior's Toy Hunt, where learners finish lessons to collect puzzle pieces for avatar outfits. He said such events "can be a meaningful opportunity to grow in-app purchases over time", but the quarter that held the first one still shows purchases falling by more than a third.
What to watch: advertising and in-app purchase revenue in the fourth-quarter table. Growth there would back Munson's claim; another fall would leave subscriptions as the only engine while the free audience keeps expanding.
Duolingo Ends Max Without Showing Its Size
Duolingo Max, the most expensive tier, existed mainly to sell unlimited Video Call. For three quarters von Ahn said he had not decided its future once Video Call moved into the cheaper Super plan. He now has: "we've decided to sunset Duolingo Max, our highest subscription tier".
This month the app stops selling Max and moves individual Max subscribers to a Super family plan at the same price. Von Ahn wrote that "The bookings effect of this should be minimal"; family-plan Max subscribers get no answer until 2027.
Nobody outside the company can check that claim, because it last disclosed Max's share of subscribers a year ago, at 9%, and never said what bookings those users carry. Each moved subscriber gets a family plan for the Max price, so Duolingo forgoes the gap between the two prices.
What to watch: subscription bookings growth in the fourth quarter against this quarter's rate. Holding near it with Max gone would confirm the effect is small; a drop toward 10% would mean Max was worth more than von Ahn let on.
Risks This Quarter Didn't Answer
Six standing concerns got no new numbers in the shareholder letter, and one deepened: the Eggy Party tie-in adds users in China, a market where the law requires local AI models and regulators set the terms the company operates under.
- China: no regional user figures; the next filing's geographic note is the only place to size the exposure.
- App-store dependence: no update on direct web payments that would cut the fees Apple and Google take; the next filing's concentration note will show their revenue share.
- Retention: the current-user retention rate, disclosed at a record 84% last quarter, does not appear in this letter; its return in February would end the question.
- AI substitution: the letter says nothing about chatbots or machine translation as rivals for a second quarter; a user-growth slowdown in any region would reopen it.
- Founder equity: no update on the modified founder award that cut stock-pay expense last quarter; the next filing's equity note will show the charge.
- Key person: no change; von Ahn's public comments remain the one lever that moved user growth both ways in 2025.
👓 Other Themes
Bonus for the User Target Trims Profit
Duolingo booked an $11.4 million charge for a company-wide bonus that pays out if fourth-quarter daily user growth reaches 25%. Munson said the bonus "was previously excluded from our outlook", and the fourth-quarter plan now carries the rest of it. By our math, adjusted EBITDA without the charge would be about $90 million, a 29% margin, so the reported drop to 25.5% is mostly the bonus.
Currency Flatters the Year, Hurts the Fourth Quarter
Duolingo sells in many currencies, so the dollar's moves change its reported numbers. For 2026 the company expects bookings growth of 12% as reported but 11% at constant exchange rates, so the weaker dollar adds about a point. For the fourth quarter the effect reverses, with 13% reported against 14% at constant rates; each 1% move in the dollar shifts quarterly bookings by about $2 million, Munson said.
💲 Other KPIs
Free cash flow fell 3% from a year ago to $75.3 million even though operating cash flow rose. The gap is capital spending, which doubled to $14.1 million by our math as Duolingo buys equipment and builds software. Cash margins have narrowed for two quarters after a seasonal high in the first quarter; nine-month free cash flow stands at $301.8 million by our math.
Buybacks slowed sharply, to $6.4 million in the quarter from $44.4 million in the spring. Through October 30 the program had spent $82.0 million, leaving $318 million of the authorisation by our math. Munson said the purchases offset all dilution from the previous two years, yet the diluted share count still rose 2% from a year ago, to 50.8 million.
Deferred revenue, cash customers have paid for lessons not yet delivered, reached $513.6 million, up 16% from a year ago. The balance has barely moved since March, by our math, because bookings growth stalled in the spring; a fourth quarter on the bookings plan would refill it and feed next year's revenue.
🔮 Guidance
New. The plan has bookings growing 13% in the fourth quarter, about $2 million above what the August full-year outlook implied for the period by our math, which von Ahn described as slightly ahead. The comparison is hard, because the year-ago quarter itself grew 24%. Munson pointed to the Q3 user outperformance and to advertising and in-app purchases.
Raised. The target moved up $11 million, to about 12% growth, the high end of the 10–12% range set in February. By our math the fourth quarter carries the whole increase, since nine months are already booked. It is the second raise in a row after a cut in May: −$6M, +$5M, +$11M, each move larger than the last.
New. The plan has revenue growing 14% in the fourth quarter, roughly matching bookings growth for the first time this year, because the strong bookings of late 2025 now reach the revenue line. By our math the full-year outlook leaves $321.6 million for the quarter, consistent with the point figure. At constant currency the growth would be 13%.
Raised. The target moved up $14 million, to about 18% growth, the top of the 15–18% range set in February. The moves this year: −$4M in May, +$2M in August, +$14M now. Most of the year is already locked in, because Duolingo spreads subscription revenue over twelve months; the fourth quarter is the only open piece.
New. Adjusted EBITDA is profit before interest, tax, depreciation and stock pay. By our math the fourth-quarter figure sits $0.4 million below the same quarter last year on 14% more revenue, because the outlook now includes the rest of the staff bonus tied to user growth and some costs that slipped from Q3 into Q4.
Raised. The full-year target moved up $3 million and now includes the user-growth bonus that earlier plans left out, so the underlying raise is larger than it looks. Munson noted the figure sits $21 million above the original February target. The moves this year: +$8M, +$10M, +$3M.
Raised. The plan has gross margin at 72.0% in the fourth quarter, down 0.9 points from Q3 as Video Call reaches all Super subscribers, but three points above the 69% Duolingo set in February. By our math the full year would land near 72.6%, against the 71.6% the company expected in August.
New. Von Ahn expects fourth-quarter daily user growth "similar to Q3", which by our math means about 68 million daily users against 52.7 million a year ago. That would clear the 25% threshold that triggers the company-wide bonus, which is why the bonus now sits in the profit plan.
Lowered. Stock pay should run about 14% of revenue for the year, from around 15%. Nine-month stock pay was 13.4% of revenue by our math, so the figure implies about $51 million in the fourth quarter, the largest quarterly amount of the year.
Unchanged. Equity grants should add 3.5% to 4.0% to the share count before repurchases, the same range the company has carried all year.
Unchanged. The rate ran 23.6% this quarter by our math, inside the range, after rising from the 18–20% framed in May.
❓ Key Questions
How many daily users did the game tie-ins bring?
Von Ahn says Brawl Stars and Eggy Party brought millions of new and returning users but gives no figure, so the lasting part of the 29% growth cannot be separated from the campaign part.
What share of subscribers and bookings did Max carry?
Duolingo last disclosed Max at 9% of subscribers a year ago. Without the current share and its bookings, nobody can test the claim that ending Max costs little.
Does the $375 million free cash flow floor still stand?
The August outlook had free cash flow above $375 million. This letter raises every other full-year line but does not mention cash; nine-month free cash flow is $301.8 million by our math.
Why did in-app purchase revenue fall 38%?
Purchases of virtual goods dropped to $6.9 million while the fourth-quarter outlook cites strength in that line. The cause, whether fewer buyers, the longer trials or a product change, is not explained.
What is the current-user retention rate now?
Duolingo called 84% an all-time high last quarter and does not repeat the figure here, although retention is the stated driver of daily user growth.