The Funnel Finally Widened: Duolingo's Users Accelerated, Its Profit Guide Went Up Again, and the Stock Fell 9%

Results vs. Consensus

Duolingo reported after the close on August 5. Two quarters into an investment year that reset the equity in February, the useful question remains whether management is delivering what it promised. This quarter it delivered more than it promised on every financial line and, for the first time, produced evidence on the user question that the whole strategy rests on.

Year-over-Year Comparison

Sequential Comparison

Users, and the answer to last quarter's question

In May we wrote that the entire Q1 user result came from retention, that the CEO had said the top of the funnel was about flat, and that if it was still flat at the third-quarter print the 2028 target and the rating would both be at risk. That question was answered a quarter early and in the affirmative.

Daily users grew 23% to 58.7M, accelerating two points from Q1 and running ahead of the roughly 20% framework. Monthly users grew 10% to 140.6M, up from 5.8% growth in Q1 and 2.0% sequentially. Management attributed the acceleration to three factors, of which it described two as permanent: product changes and marketing impact, plus a one-time Streak Revival campaign in June. On that basis it now expects daily-user growth above 20% for the remainder of the year, an upgrade from the flat 20% framework carried since February.

Retention improved alongside acquisition rather than instead of it. The company disclosed Current User Retention Rate, the proportion of engaged daily users who return the next day, at an all-time high of 84%, up about one point year over year. The daily-to-monthly engagement ratio reached 41.8%, up 460 basis points year over year and 80 basis points sequentially, and has now risen in every quarter we have tracked.

The one user metric that decelerated is paid subscribers, at 12.7M and 17% growth against 21% in Q1. That is the intended consequence of pulling back on conversion friction, and it is the mechanism by which the strategy costs money.

Segment Performance

Duolingo reports as a single segment. Two decompositions carry information this quarter: the revenues-by-product-type appendix, and the growth-driver detail, where management gave more disclosure than in any prior quarter we have covered.

This table is the sharpest check available on the strategy, and it does not flatter it. Subscription grew 22% and is carrying the whole result. The two streams that should benefit most from a deliberately larger free base did not: advertising grew 2%, down from 15% in Q1, and in-app purchases fell 23% as the energy mechanic laps its rollout and as friction is removed from the free tier. The English Test was flat. Non-subscription revenue in aggregate is roughly unchanged year over year while the free base grows, which is the strongest available argument that the monetization replacement levers management describes are still ahead of it rather than underway.

The Green Machine and retention

Management named its internal experimentation process for the first time and quantified its cadence: a new version of the app every week, each carrying roughly 350 changes, tested and kept on measured impact. The output metric it now discloses is Current User Retention Rate.

"We're very happy that it is an all-time high and that it has increased by about a percentage point in the last year because tiny changes to CURR end up implying pretty large changes in daily active users over time because it really compounds. This is pretty broad-based. Basically, you see it in every region, you see across all types of users." — Luis von Ahn, Co-Founder and CEO

Assessment: Disclosing a named retention metric with a level and a trend is a real improvement in transparency and gives the market its first quantitative handle on the product half of the strategy. A one-point annual gain in next-day retention sounds trivial and is not: compounded daily across a base this size it is the mechanism management is relying on, and it is the one thing here that is fully within its control.

Marketing, and the shift off owned social

The acquisition channel that broke in 2025, the company's own viral social accounts, has been supplemented rather than repaired. Owned accounts still generate over a billion organic impressions per quarter, but the company now works with hundreds of creators worldwide, and in China, Indonesia and India roughly two thirds of social impressions come from influencer-generated content. Separately, performance marketing channels and creative output both tripled over the past year.

"What we're finding is that we are able to not spend a ton of money on this and what's nice is that these people have different audiences than us. And so this really just brings in a lot of new users." — Luis von Ahn, Co-Founder and CEO

The CFO added that top-of-funnel growth improved in almost every region during the quarter.

Assessment: This is the structural fix to the specific fragility we flagged at initiation, that user acquisition depended on the virality of one social voice the company had demonstrated it could turn on and off. A creator network with different audiences plus a real performance-marketing function is a diversified funnel. It is also cheap: management was explicit that the influencer approach does not require heavy spend.

Streak Revival

A one-time June campaign let learners restore their longest-ever streak by opting in and completing three lessons. It reached 15.4M learners, including nearly 8M who had no active streak when it began.

"More than 15 million learners revived their streaks. And what's particularly encouraging is that these users are also showing better retention than a typical reengaged cohort." — Luis von Ahn, Co-Founder and CEO

Assessment: Management flagged this as the one non-recurring component of the acceleration, which is the right disclosure and should be treated as a partial discount on the 23%. The more durable read is what it demonstrates about the dormant base: nearly 8M lapsed users returned for a streak, which sizes a reactivation pool the company has barely touched. Better-than-typical retention on a reengaged cohort is the detail that makes it more than a stunt.

Math and Music

Both subjects sit at single-digit millions of daily users, materially smaller than chess, and management was blunt that neither moves the aggregate near-term. The more useful disclosure was a strategic correction on Math.

"When we launched Math, I thought -- and I was wrong, I thought that we could get the average person on the street to get addicted to learning math. I was wrong. I don't believe we can do that. But -- so now our strategy is actually to mainly teach math to the people who actually need to learn math, which are basically K through 12 students." — Luis von Ahn, Co-Founder and CEO

Assessment: A CEO stating plainly that a prior strategic assumption was wrong is worth more than most positive disclosures. The K-12 repositioning is also the right one, and it matches the parental-willingness-to-pay observation from February. Neither subject is a 2026 or 2027 number; both are options.

Key Topics & Management Commentary

Overall Management Tone: Confident and noticeably more specific than the previous two calls, with new named metrics, a named internal process and a disclosed internal bonus trigger that no one asked about. Management volunteered the one-time component of its own good news and separated it from the durable components. The least settled area was the premium tier, where the honest answer to what happens to Max was that it might be discontinued and the decision has not been made.

1. The funnel widened, and management said so directly

Monthly-user growth went from 5.8% to 10% and daily-user growth accelerated for the first time since the strategy began. Both the CEO and CFO were asked about the top of the funnel and both confirmed the improvement without hedging.

"One thing I might add to that, Andrew, is the top of the funnel is part of the story of the quarter. It got better. As you know, that's been a focus area of ours. We've talked about how we want to get better there. And in almost every region, the rate of growth there improved in top of the funnel in the quarter, which is a great accomplishment for the team." — Gillian Munson, Chief Financial Officer

Management also noted that United States growth increased notably and that all regions are growing faster than before, with Asia still the fastest.

Assessment: This is the single most important fact in the quarter and it is the specific test we set in May. Region-broad improvement is a stronger signal than an aggregate number, because it makes a one-market or one-campaign explanation harder to sustain. Discount the 23% for Streak Revival and the underlying acceleration is smaller but still real.

2. Video Call now costs under a cent

The most consequential disclosure of the call was a unit-cost figure. At launch, a Video Call session cost roughly thirty cents to serve, which is why the feature was placed behind the most expensive tier. It now costs under one cent, achieved mainly by moving to open-weight models with no observed quality loss.

"The good news is that through a lot of really hard work, we've been able to bring down the cost of video call. It is now under $0.01 per video call. And the reason for that is mainly a move towards open source models. It's just a lot cheaper to do that, and we don't see a loss in quality in there." — Luis von Ahn, Co-Founder and CEO

Most new Super subscribers now receive Video Call, with existing Super subscribers expected to gain access over the coming months.

Assessment: A more than 96% reduction in the unit cost of the company's most expensive and most differentiating feature is the clearest quantification anyone has given of what falling inference costs are worth to a consumer AI business. It explains three consecutive gross-margin beats, it is why the tier change is affordable, and it converts the AI cost line from a structural bear point into a structural tailwind.

3. Max may be discontinued

With Video Call moving down a tier, the top tier's reason to exist is in question, and management said so.

"One possibility could be that, while Super subscribers get a limited version of video call, like limited number of video calls and Max subscribers get unlimited. That is a possibility. Another possibility truthfully is that we may actually sunset Max. But -- what I will tell you is that we're going to have an answer to this in the next couple of quarters." — Luis von Ahn, Co-Founder and CEO

Management added that the deliberate pace is precisely to avoid revenue loss in the transition.

Assessment: Openly contemplating the retirement of a premium tier that was 9% of subscribers a year ago is a real disclosure with a real revenue risk attached. It is also consistent: if the cost that justified the tier has fallen 96%, the tier's economic rationale has gone with it. The commitment to resolve within two quarters puts a date on it, and the stated intent to do it without losing revenue is the constraint to hold management to.

4. Super Lite, an ad-supported tier in test

A new tier is being tested at roughly half the price of Super, ad-supported, offering double energy rather than unlimited. Management was emphatic that it is a test and that only a small fraction of subscribers are on it, partly because it is barely being promoted.

"It is cheaper than Super, it's about half the price, depending on the geography. It is ad supported, so you get ads. And in addition to that, basically, you don't get unlimited energy, you get twice as much energy... the goal with a Lite plan like this would be, if we're pretty sure that you're not going to buy Super then we should try to sell you Super Lite." — Luis von Ahn, Co-Founder and CEO

Assessment: The clearest example yet of monetization that is aligned with rather than opposed to user growth. A cheaper ad-supported tier converts free users who were never going to buy Super, so it adds revenue without adding friction to anyone. The risk is downtrading from Super, which is why the price gap and the energy allowance matter, and neither was disclosed with enough precision to model.

5. Longer free trials are working

The one non-friction monetization lever identified in May has scaled. Most free trials are moving from the historical seven days to one month, with a mechanism management explained in full.

"Significantly more people decide to take that trial because it's just a better deal. And because of that, we got a larger number of people actually converting to payers. And the other nice thing is that as soon as they agree to go on the free trial, the experience just gets better because we turn off energy, we turn off the ads. So this actually increases daily active users as well." — Luis von Ahn, Co-Founder and CEO

Assessment: A lever that raises trial uptake, raises conversion and raises daily users simultaneously is exactly the category the February strategy said existed and could not yet name. It also compounds with the acquisition improvement, since a wider funnel feeds more trials. This is the most under-discussed positive in the release.

6. The profit outlook went up again

Full-year Adjusted EBITDA guidance rose to approximately $320M on a roughly 26.5% margin, from $310M at the Q1 call and $299M to $305M when the year was framed in February. Full-year free cash flow guidance rose to over $375M from over $350M. Gross margin guidance improved again, to approximately 71.0% in Q3 and approximately 71.6% for the year, against the roughly 69% exit rate contemplated in February.

"We are raising our full-year Adjusted EBITDA margin outlook by nearly one point to approximately 26.5%, reflecting flow through from stronger-than-expected gross margin performance." — Gillian Munson, Chief Financial Officer

Assessment: Two raises in two quarters, totalling $15M above the original range's high end, on a plan whose entire premise was accepting lower profitability. The investment year is costing meaningfully less than the February guide assumed, and the source is a cost curve rather than under-spending, which means it is not being bought at the expense of the strategy.

7. A disclosed internal bonus tied to Q4 users

The CFO volunteered, unprompted and not included in guidance, that the company has a bonus plan triggering if Q4 daily-user growth reaches 25% or higher, with a cash payout of roughly $10M in Q1 2027, potentially more if growth is higher.

The disclosure does two things at once: it flags a possible unguided expense, and it reveals that management's internal target sits two points above the external framework it just raised to "above 20%."

Assessment: A rare and useful piece of information. The company just told the market what number its own incentives are pointed at, and it is 25%, not 20%. It also handled the accounting honestly by excluding an uncertain payout from guidance rather than burying it. If Q4 daily-user growth prints at 25%, the roughly $10M expense will be the best money the company spends this year.

8. Open-weight models as an explicit strategy

Beyond Video Call, management framed the model portfolio as a deliberate and shifting mix, with a stated preference for open-weight wherever quality permits.

"At the highest level, our view inside the company is that it is in our best interest as a company to use open-weight models as much as possible. So if I had a magic wand, I would try to move everything to an open-weight model. It's not always possible because sometimes the frontier models are more advanced. But from a company standpoint, it is just significantly better because it's way cheaper to use open-weight models." — Luis von Ahn, Co-Founder and CEO

Management sized the AI bill for the first time: tens of millions of dollars in cost of goods sold, alongside hosting as the other major component, plus roughly $10M of internal AI usage.

Assessment: The first hard sizing of the AI cost base in four quarters of coverage, and it is smaller relative to a roughly $1.2B revenue base than the 2025 bear case assumed. The portfolio approach, with a stated intent to weight further toward open-weight models over time, means the cost line should keep falling on mix alone even if frontier pricing does not.

9. Advertising is being taken seriously

Management described a shift in its own view of advertising on both the revenue and spend sides, with an unusually candid account of how little the business had previously invested.

"There was a time at Duolingo a couple of years ago, where our entire investment in ad revenue was half a person, there's one person whose halftime job was to deal with ads. So this is how much we were investing in them. We now have a team it's significantly much more professional." — Luis von Ahn, Co-Founder and CEO

Management said it expects visible improvement in the advertising business over the next few quarters, while stressing that subscription will remain far larger for the foreseeable future.

Assessment: Advertising is the monetization vector most aligned with a strategy that grows the free base, and management is now treating it as a business rather than an afterthought. The comparison it drew, that apps of similar or larger scale make far more from advertising, is the size of the gap. Set against the commentary, though, the disclosed line grew 2% this quarter against 15% in Q1, so the professionalization is not yet visible in the revenue. This is an option, and the option has not started paying.

10. China becomes the largest market, on local models

China is now the second largest market by daily users and management expects it to become the largest within a year or two, monetizing about as well as France. All AI usage in China runs on Chinese models by law, and management acknowledged the regulatory exposure without minimizing it.

"It is our second largest market in terms of daily active users, and we expect it to become our largest market in terms of daily active users in a year or 2... In terms of regulatory risk, it's important to mention that, of course, we understand that China poses some regulatory risk. That's kind of outside of our control, how they operate the country." — Luis von Ahn, Co-Founder and CEO

Assessment: A market on track to be the largest by users, monetizing near a developed-Europe rate, is materially more important to the story than it was two quarters ago, and the concentration risk rises with it. The disclosure that the AI stack in China is entirely local is new and is the right architecture, since it decouples the China product from any restriction on US model access.

11. The monetization posture is loosening again

Asked whether the emphasis is shifting back toward monetization, management conceded the point and dated the change.

"It is true that our experimentation, when we started the year, we had a little bit of a pause in monetization in the sense that we're like whoa, whoa, whoa, wait, any monetization that we're doing that is counter to DAU growth, don't do it... At this point, enough time has passed that we understand levers significantly better. So we are a little more back into like, okay, this type of monetization is good." — Luis von Ahn, Co-Founder and CEO

Management explicitly rejected the idea of a single future moment when the company switches from users back to monetization, describing instead a gradual increase in monetization focus as aligned levers are found.

Assessment: The most important forward statement on the call for anyone modelling 2027. The February framing implied a discrete investment period followed by a return; management is now describing a continuous process in which aligned monetization scales alongside users. That is a better business but a harder one to model, and it means the bookings growth rate recovers gradually rather than snapping back.

Guidance & Outlook

The revision pattern of the prior quarter repeated and widened. Profit guidance went up on every line: Adjusted EBITDA by $10M, the margin by 80 basis points, free cash flow by $25M, and the gross-margin path against the February trajectory, where the like-for-like comparison is the year-end exit rate management now puts closer to 70% against the roughly 69% it first expected. The top-line ranges were reaffirmed and the point estimates nudged up slightly. Users, the metric management cares most about, were guided up from approximately 20% to above 20%. The only guidance line to move against shareholders was the tax rate, which rose for the second consecutive quarter.

Implied Q-over-Q ramp: The $1,285M full-year bookings estimate less the $308.5M and $289.1M already reported and the $307M Q3 guide leaves roughly $380M for Q4, implying growth near 13%, again the fastest quarter of the year. On revenue, the $1,207M full-year estimate less the three known quarters leaves roughly $315M for Q4, implying growth near 11%, in line with Q3. The bookings shape is therefore back-loaded and the revenue shape is not, which is what you would expect if new users are arriving now and monetizing on a lag.

Street at: Q2 revenue consensus was $297.3M against $298.5M delivered, and EPS $0.61 against $0.66. The Q3 revenue guide of $302M is 1.3% below the $305.9M consensus, and the full-year revenue guide is broadly in line. There was no clean two-source pre-print Adjusted EBITDA consensus, but a Street carrying the prior $310M full-year figure received $320M.

Guidance style: Still conservative on profit and now honest about it. Adjusted EBITDA has beaten its quarterly guide by 13.3% and then 8.9% in the two quarters since the CFO said she intended to narrow the gap between guidance and actuals. The gap is narrowing but remains wide. The bookings and revenue guides have been within a couple of points both quarters, which is the intended behavior; the profit guide is where the conservatism now lives.

Analyst Q&A Highlights

Why faster users did not produce a bigger bookings raise

The opening question posed the central tension of the quarter directly: if daily users are running above plan, why is the bookings guide barely moving? The answer restated the operating framework rather than defending the arithmetic.

Q: "Given DAUs are now expected to be above that 20% year-over-year growth in the second half of the year, could you maybe give some color as to why the full year bookings guide wasn't raised more? Like I realize you're in experimentation mode, but shouldn't a larger amount of users on the platform technically translate to increased bookings even if you're not pushing for monetization."
— Wyatt Swanson, D.A. Davidson & Company

A: "The first thing to note is that our users don't monetize immediately... it takes a while for them to monetize because we have this premium model. We do expect that higher DAUs will imply higher revenue. But it's going to take some time. And the second reason is exactly what you said. We said at the beginning of this year that we were going to operate in this box in terms of revenue, and we are going to continue with this box."
— Luis von Ahn, Co-Founder and CEO

Assessment: The honest answer, and the one that explains why the stock fell on a good quarter. Management is choosing not to convert the user upside into near-term bookings, which is the strategy working as designed and is also exactly what a shareholder focused on the next two quarters does not want to hear. The lag argument is real: subscription conversion follows acquisition by quarters, not weeks.

The top of the funnel

The question we flagged in May as the one that decides the thesis was asked again, and this time the answer had changed.

Q: "As we think about the U.S., Luis, understood the strength in the quarter and kind of resurrected users. Can you talk about top of funnel though? How do you feel about trends there and kind of the broader opportunity of attracting new users that may be new to Duolingo?"
— Andrew Boone, Citizens JMP

A: "While still the majority of our growth is organic, we're getting good traction on both usage of influencers and performance marketing. And we are seeing that in top of funnel. So it is increasing. So we actually feel pretty good about our top of funnel. It is increasing. And you asked about the U.S., in particular, the U.S. growth has increased quite a bit in the last quarter."
— Luis von Ahn, Co-Founder and CEO

Assessment: One quarter after saying the top of the funnel was about flat, the CEO says it is increasing, and the CFO independently confirmed improvement in almost every region. That is the specific reversal the entire 2026 plan was built to produce, arriving a quarter earlier than we expected to be able to grade it.

What happens to the premium tier

With Video Call reaching Super subscribers, an analyst asked what the learnings mean for Max. The answer was more open-ended than any prior quarter's.

Q: "Interested to hear with the learnings you've seen so far from giving video call to new Super subscribers? And how is that influencing your plans for video call and Max generally as we go from here?"
— Nathaniel Feather, Morgan Stanley

A: "Now that calls into question what are we doing with Max? And my answer is, I don't know yet... Another possibility truthfully is that we may actually sunset Max. But -- what I will tell you is that we're going to have an answer to this in the next couple of quarters. And in addition to that, we're going to try to do this without a loss of revenue."
— Luis von Ahn, Co-Founder and CEO

Assessment: Naming tier retirement as a live possibility on an earnings call is unusually direct and carries a real revenue risk. The mitigating facts are the stated two-quarter deadline and the explicit revenue-neutrality constraint. Anyone modelling Max as a durable premium contributor into 2027 should stop.

Retention across cohorts

A question on the composition of the record retention figure asked whether it was concentrated in particular cohorts. The answer attributed it to accumulated product work rather than any single change.

Q: "Good to see CURR at an all-time high. Just curious, could you share more color on just overall retention by cohort. Are these new free trial users that are engaging on a daily basis, exhibiting higher engagement trends?"
— Bryan Smilek, JPMorgan

A: "It's hard to pinpoint to a single one because again, every single -- we have a new version of the app every single week and every version of the app has approximately 350 changes. So it's hard to point to something that actually did it. But it's just generally our product is just stickier and we're very happy with that because that's kind of the best type of growth you can expect that is just pure, it's a better product."
— Luis von Ahn, Co-Founder and CEO

Assessment: An honest non-answer that is more reassuring than a clean attribution would be. Retention gains that cannot be traced to one feature are harder for a competitor to copy and less likely to reverse when a single mechanic is lapped. The 350-changes-per-week cadence is also the first real quantification of the experimentation engine.

Whether the emphasis is shifting back to monetization

An analyst read the call's texture correctly and asked whether the balance is tilting back. Management agreed, with an important qualification about how the transition happens.

Q: "As I listened to the call and what we're talking about here, feels a little bit like we're starting to maybe shift or start to focus a little bit more on monetization, talking about ads and new tiers. Is that fair, not to say that you're moving away from the optimization around DAUs, but it does feel like you are maybe shifting a little bit the monetization side. How do you think about that transition? When is the right time?"
— Ygal Arounian, Wedbush

A: "A lot of people ask us, 'Oh, when are you going to stop worrying about DAUs and when are you going to start making money?' A lot of people ask us that. I don't think that we're going to have a shift like that. we're going to continue really trying to get to 100 million DAUs in 2028. But you'll see probably increased monetization focus kind of gradually over time as we find ways to monetize that are not at odds with DAU growth."
— Luis von Ahn, Co-Founder and CEO

Assessment: This reframes the whole investment case. February implied a discrete investment period with a return date; management now describes a continuous ratchet with no switch to flip. That is the better long-term construction and the worse one for anyone waiting to buy a growth reacceleration event. It also means bookings growth recovers gradually rather than stepping.

Open-weight models and the margin structure

A question on the durability of the AI cost savings produced the clearest description of the model strategy and its limits.

Q: "I want to come back to the comments you made about the open source models and how that's driven down AI compute cost. So, a, how should we think about that opportunity in terms of the margin structure over time?"
— Ygal Arounian, Wedbush

A: "It just turns out that for a lot of applications, you don't need the absolute smartest model, like the reality is the quality is indistinguishable for many applications that we use. For example, serving a call like a conversation practice to somebody who can only speak 100 words of a language. You don't need to serve a -- your model doesn't need to be as good as a philosopher."
— Luis von Ahn, Co-Founder and CEO

Assessment: The reasoning is sound and specific to this workload: teaching a beginner is a low-capability task that a cheap model serves as well as an expensive one. That is a durable structural advantage over consumer AI products whose value depends on frontier capability, and it is why this company's inference cost curve should keep falling faster than the category's.

Confidence in the fourth-quarter acceleration

The final guidance question asked what underwrites the implied Q4 step-up. The answer was that the year is playing out as planned rather than that something new has been observed.

Q: "Could you talk through just what's driving the confidence level behind that acceleration in Q4? Is that simply easy comps? Or are you just seeing something with the product side finally starting to click there."
— Justin Patterson, KeyBanc

A: "The reality is how the year is playing out hasn't really changed too much. Of course, we've held the guidance. Q2 was a little better than we expected... I think it's important to remember that our bookings guidance also includes about 0.5 point of FX headwind. So net-net, you're probably at a little bit of a better place. But I think the year is playing out the way we thought."
— Gillian Munson, Chief Financial Officer

Assessment: The half-point currency headwind is the detail worth extracting: the reaffirmed bookings guide is slightly better on an underlying basis than the headline suggests. The rest is a statement that nothing has changed, which is neither reassuring nor alarming. The Q4 step-up is still the least-evidenced part of the plan.

What They're NOT Saying

1. How much of the daily-user acceleration was Streak Revival: management named the campaign as one of three drivers and said two are permanent, but gave no split. With 15.4M learners involved against a 58.7M base, the campaign could account for a large share of the two-point acceleration or a small one, and the difference decides whether the trend is real.
2. Why the full-year tax assumption rose for the second consecutive quarter: from approximately 6% benefit in February, to 18% to 20% in May, to 23% to 25% now. Three different assumptions in three quarters, none explained, and no analyst has asked about any of them.
3. Chess: disclosed at approximately 7 million daily users in February and not updated since, for a second consecutive quarter, despite being described as the company's most successful new subject and receiving dedicated marketing.
4. Any Max disclosure while its future is being decided: no subscriber share, renewal rate or revenue contribution, in the quarter management said the tier might be discontinued. The revenue at risk from a sunset is therefore unquantifiable from outside.
5. Super Lite economics: price described only as "about half" of Super and varying by geography, energy allowance as "twice as much," and penetration as a small fraction. None of it is precise enough to model, in a tier that could reprice a meaningful part of the subscriber base.
6. Why advertising decelerated while management talked it up: the appendix puts advertising revenue at $21.1M, up just 2% year over year against 15% in Q1, in the same quarter management devoted substantial airtime to the opportunity and to a professionalized team. Neither the letter nor the call reconciles the commentary with the number, and no share-of-bookings figure was given.
7. What the paid-subscriber deceleration costs: subscriber growth slowed from 21% to 17%, and no ARPU figure was given for the second consecutive quarter, so the split between fewer subscribers and price or mix effects cannot be determined.
8. Dilution against the buyback: estimated diluted shares rose 2.0% sequentially to 50.7M despite roughly $44M of repurchases in the quarter, with RSUs outstanding the driver. Management repeated the 3.5% to 4.0% dilution guidance without addressing whether the buyback is keeping pace on a net basis.
9. The Duolingo English Test, on the call: a fourth consecutive quarter with no mention by management or any analyst, in a quarter where the appendix shows it flat year over year at $10.1M. The credentialing ambition set out three quarters ago has gone unaddressed since.
10. Whether the Q4 bonus threshold is achievable: management disclosed a 25% Q4 daily-user trigger and said it is currently uncertain whether it will be met, with no indication of where quarter-to-date growth is running, in a call that otherwise offered quarter-to-date color in prior quarters.

Market Reaction

• Pre-print setup: DUOL closed at $135.32 on August 5, down 22.9% year to date against the S&P 500's +12.8% and down 60.2% over the trailing twelve months, but up 4.3% over the trailing thirty days and 34% above the February reaction close of $101.00. The 52-week closing range entering the print was $90.03 to $390.84.
• Next-day session: the stock opened at $130.11 on August 6, 3.9% below the prior close, traded a range of $112.02 to $135.14, and closed at $122.58, down 9.4% or $12.74.
• Intraday pattern: the low of $112.02 marked a 17.2% decline, so the stock recovered 9.4% off its low, though it still closed just below the midpoint of the day's range.
• Volume: 3.8M shares against a 30-day average of 1.2M, or 3.1x normal, the lightest of the four reaction days we have covered.
• Market context: the S&P 500 fell 0.2% on the session, so essentially the entire move was idiosyncratic.

The shape is the same as May and larger: a modest opening gap, a sharp intraday flush, and a substantial recovery into the close. The stock finished 9.4% above its low, which means the $112 level found real demand. Across four consecutive reaction sessions the volume multiple has fallen from 10.1x to 8.1x to 3.6x to 3.1x, describing a shareholder base that has finished turning over and a float that no longer produces forced selling on a guide-driven disappointment.

The trigger is identifiable and narrow. The Q3 revenue guide of $302M sits 1.3% below consensus and implies revenue growth decelerating from 18% to 11%, which is the sharpest single-quarter step-down in the guided plan. Everything else in the release was better: users accelerated, the funnel widened, gross margin beat again, the profit outlook was raised twice over, and free cash flow guidance went up. The market discounted all of it against one forward revenue line.

That is a defensible trade for a fund that owns this stock on next-twelve-month revenue and an indefensible one for a fund that owns it on the 2028 case, which is the case management has been arguing for three consecutive quarters. The reaction says the marginal holder is still the former. The declining volume multiple says there are fewer of them each quarter.

Street Perspective

Debate: Is the user acceleration real or a one-time campaign?

Bull view: The bull case being made is that management itself separated the drivers and called two of the three permanent, that monthly-user growth nearly doubled to 10%, that the improvement was broad across almost every region including a notable pickup in the United States, and that next-day retention hit an all-time high independently of any campaign. A one-time streak promotion does not move regional top-of-funnel rates.

Bear view: The bear camp contends that 15.4 million streak revivals against a 58.7 million daily base is a very large number, that nearly 8 million of those users had no active streak at all, and that management conspicuously declined to size the campaign's contribution to the two-point acceleration. The guide for above-20% growth for the rest of the year rests on an unquantified split.

Our take: The bulls have it, on the strength of the monthly-user line rather than the daily one. Streak Revival reactivated lapsed users, which lifts daily users directly; it does not obviously lift monthly-user growth from 5.8% to 10%, nor regional top-of-funnel rates, nor next-day retention. The refusal to size the campaign is a real disclosure failure and the reason this stays a debate rather than a settled question.

Debate: Does a 96% cut in Video Call cost change the business or just the margin?

Bull view: The optimistic reading is that it changes the business. A feature that cost thirty cents a session had to sit behind a premium tier; at under a cent it can be given to every subscriber and eventually every user, which converts the company's most differentiating capability from a monetization gate into an acquisition and retention asset. It also explains three consecutive gross-margin beats and two profit-guide raises.

Bear view: The skeptics note that management said plainly that Video Call is one of the main things that gets people to buy, so universalizing it removes a purchase driver, and that the company has not found the replacement. Cheaper delivery of a feature you then give away is a margin benefit and a revenue risk at the same time.

Our take: Both are right and the bulls are more right. The bear point is real and management conceded it unprompted, which is why the rollout is deliberately slow and why the Max decision is still open. But a company whose flagship AI feature costs a cent to serve has structurally different economics from one paying frontier rates, and that advantage compounds with every additional feature. The purchase-driver problem is solvable; a permanent inference cost disadvantage would not have been.

Debate: Is the valuation still cheap after a 21% rally off the February low?

Bull view: The bull argument is that the enterprise value is roughly $4.9B against guided 2026 free cash flow above $375M, which is about 13x, for a business with no debt, 21% of its market value in net cash, daily users accelerating, a raised profit outlook and a 2028 target management is now incentivized against internally.

Bear view: The bear response is that the enterprise value has risen roughly a quarter since February while the top-line guide has barely moved, that reported net income fell 26% and free cash flow fell 9%, and that diluted share count rose 2% in a quarter with a $44M buyback, so per-share economics are being diluted faster than the repurchase offsets.

Our take: The dilution objection is the strongest bear point in this note and it has not improved. But the multiple has re-rated on an improving fact set, not on nothing: since February the company has raised profit guidance twice, raised free-cash-flow guidance, raised the user growth outlook, and demonstrated the funnel widening. Roughly 13x free cash flow for that combination is not demanding, and 3.8x bookings still prices low-teens growth as permanent.

Model Update Needed

Valuation impact: At the $122.58 reaction close and 50.7M estimated diluted shares, the equity is worth roughly $6.2B. Netting $1.31B of cash and short-term investments gives an enterprise value near $4.9B, or approximately 3.8x the full-year bookings estimate, 4.1x revenue, 15.3x Adjusted EBITDA and roughly 13x guided free cash flow. We raise the target to $155 from $140, which applies approximately 5.1x to full-year bookings and 20.5x to Adjusted EBITDA, and implies +26% from the reaction close. The raise reflects the funnel evidence and two consecutive profit-guide increases, not a change in the 2028 framework.

Thesis Scorecard Post-Earnings

Overall: The thesis strengthened materially, and specifically where it needed to. Bear #1 has been the binding constraint on this rating since initiation, and this is the first quarter in which it moved in the right direction on a metric management does not fully control. Both of the other bear points were challenged again. The two reservations are that management declined to size the one-time contribution to the user acceleration, and that paid-subscriber growth is decelerating without an ARPU disclosure to interpret it against.

Action: Maintain Outperform; raise the target to $155 from $140. In May we named one condition: the funnel had to widen by the third-quarter print or the rating would go. It widened a quarter early, on a broad regional basis, alongside record retention and a second consecutive raise to the profit outlook. The stock is up 21% from where we upgraded and still trades near 3.8x forward bookings and roughly 13x free cash flow with a fifth of its market value in cash. The next thing to grade is the fourth quarter, where the bookings guide requires the year's fastest growth and where management has told us its own internal target for daily users is 25%.