Duolingo, Inc. (DUOL) Earnings Call Transcript
May 4, 2026
Earnings Call Speaker Segments
Hey everybody, good evening and thanks for joining Duolingo's webcast for the first quarter of 2026. We put out the shareholder letter for this quarter today once the market closed, and you can get a copy over on our IR site, investors.duolingo.com. Joining us today are Luis von Ahn, who co-founded the company and serves as CEO, along with our CFO, Gillian Munson; they'll each give prepared remarks, and after that we'll take questions. If you're an analyst and want to ask something, just use the raise hand option. Just so you know, we're recording this call, and right now everyone is in listen only mode. A quick note before we get going: some of what we say will be forward-looking statements about upcoming events and how the business will perform financially. Those come with risks and uncertainties that are laid out in our SEC filings, and they rest on assumptions we consider reasonable as of today; we're not taking on any duty to revise them; also, we'll be talking about financial measures on both a GAAP and non-GAAP basis. You can locate reconciliations of the two in our earnings materials, and we'd suggest you look at those while assessing how we did. With that, I'll hand things to Luis.
Appreciate that, Debbie, and thanks to everyone for being here. When it came to Q1, execution was the focus. We'd stated our intention to place priority on improved teaching along with shifts in growing users, and we carried that out precisely. DAUs rose 21% year-over-year, which matches our expectations perfectly as we undertake this strategic change. I'd like to take a few minutes to discuss what we delivered during this quarter on the language learning front, since better teaching underpins everything we are working to build. Doing speaking practice well on a mobile app has always been the toughest challenge; during this quarter, we expanded its role within the experience for both free users and paid subscribers. Spoken tokens were introduced, enabling learners to give answers verbally to nearly every exercise. We began the rollout of Speaking Adventures, placing learners into real-world conversation scenarios. Flashcards were launched, assisting learners in developing quick recall through saying words out loud. For those on paid subscriptions, Video Call continues to improve. Across the last 12 months, the average count of words spoken per user in that feature has more than doubled. A significant content milestone was also achieved. Courses reaching professional proficiency — B2 on the CEFR scale — are now available for all our nine most learned languages; reaching that point happened quickly. During Q1 by itself, we released 20,500 course units. For context, that exceeds what we shipped each quarter by over 10 times merely two years back. AI has completely transformed the realm of what's achievable for us. My belief is that we've barely begun to tap its potential. The product now stands better than at any prior point, and my excitement about what lies ahead couldn't be greater; with that, I'll hand things over to Gillian.
Thanks, Luis. The first quarter came in strong; both bookings and revenue grew by double digits, gross margin went up, and adjusted EBITDA reached $83 million — roughly 29% of revenue. When you look ahead to 2026, it makes sense to restate our thinking about that year. We're making intentional investments so that we become a bigger and more resilient business over the long haul. That means, for this year, we're steering the business toward the targets we gave on the fourth quarter call. To be specific: bookings growth of 10%-12%, revenue growth of 15%-18%, and an adjusted EBITDA margin near 25%. To assist with your modeling, we've supplied point estimates for the full year 2026, based on what's visible today and in line with those ranges. Those figures are: bookings growth of approximately 10.5%, revenue growth of approximately 16.1%, and an adjusted EBITDA margin of 25.7%. There are a few items we want to be certain are on your radar while you build out your models. On bookings, the roughly 6% growth we expect in Q2 reflects a difficult comparison. Last year's quarter contained the first rollout of Energy, a price increase on our most popular subscription plan, and unusually strong advertising performance. We do anticipate bookings growth picking up speed during the back half, with roughly three points of acceleration in Q3 and an additional increase in Q4. For revenue, we expect roughly 17% growth in Q2, and after that we anticipate growth stepping down in Q3 before leveling off in Q4. On gross margin, we're expecting approximately 71% in Q2, then a downward trend to roughly 69% by year-end as usage of AI-powered features across our products grows. Adjusted EBITDA margin in Q2 ought to be approximately 24%. We expect Q3 adjusted EBITDA margin to be flat to slightly lower versus Q2 before nearing 27% in Q4; that's a great deal to absorb, but the broad takeaway is that 2026 is a pivotal strategic investment year for us, and so far it's unfolding the way we anticipated, as shown by the point estimates for our financials that we've provided. We head into Q2 with more than $1 billion in cash, no debt, and we expect to produce more than $350 million in free cash flow this year. We intend to keep executing on our buyback authorization, under which repurchases so far total 514,000 shares, or roughly 1% of our fully diluted shares outstanding. 2026 is a major year for Duolingo, and I'm extremely excited about what we're building. With that, I'll hand it back to the operator, and we can take your questions.
Hey there, I appreciate you asking that. Thanks a lot.
Hello, Wyatt.
Sure. Could you walk us through some of the different elements driving DAU growth this quarter, and maybe also what to expect as we move into Q2? Specifically, talk about whether it's coming from performance marketing, whether word of mouth is starting to pick back up, or if there's something else at play. Also, if you could, touch on which geographies are demonstrating particular momentum or softness.
Thanks for asking. DAU growth is hugely important for us — it's clearly the number one thing we're focused on this year. We're growing in every region, just like we've been doing for multiple years now. Of course, some areas are scaling faster than others. Asia stands out as the region with the quickest expansion. On the growth side, the things pushing us forward haven't really changed much. Historically, the biggest factor has been people telling other people about us. The majority of folks who sign up for Duolingo hear about it from someone they know. We do put some money into marketing, including performance marketing. That budget has gone up somewhat, though it's still fairly small next to what apps of our scale spend. Overall, the forces driving our growth look about the same as before. To wrap up, one other area that has traditionally driven daily active user growth is simply better retention, which really comes down to making the product more engaging. That effort has paid off nicely. We've rolled out plenty of product updates—some minor, others more substantial—all aimed at lifting retention. You'd notice this through a rising DAU-to-MAU ratio, which has been climbing nearly every quarter and went up once more this quarter.
Great, thank you. Just one short additional question. Could you give us a bit of detail regarding what you anticipate for DAU growth in 2Q, and is 20% still the correct lens through which to view DAU growth as we move through 2026?
Yes. All the points we made during the previous call still stand. What I mean is, we anticipate it will remain near 20% for the entire year. There will be minor fluctuations here and there, depending on last year's comps, and so on, yet it stays around 20% for the remainder of the year. Nothing has shifted since our prior conversation.
Perfect. Thanks.
Hi, Ross.
There. Hey, everyone. I think I've got this working now. Alright. Luis, you brought up a few intriguing points 90 days ago as part of this year's plan. One involved, sort of, re-energizing engagement within the free tier — specifically that highly engaged free tier, the top 20% of it. I'd love to hear any update on those efforts. You also said you hoped to draw inspiration from the major mobile gaming companies out there, in terms of new features you might potentially bring into Duolingo. I'm curious to hear what you've learned and any fresh strategies along those lines. Thanks.
Yeah. Thanks for asking. Okay, regarding the free tier, I mean, we've essentially made it so that the free tier is better than it was two months ago. Naturally, you know, certain changes take time. Only two months have passed since the last earnings call — it's not as if we've made, you know, 1,000 changes. We have improved the free tier. There are simply more features accessible to free users. You know, we're quite pleased with that. We believe that over time this will genuinely boost word of mouth. Yeah, as for drawing inspiration from mobile games, we've always drawn plenty of inspiration from mobile games. The ultimate idea, what we're attempting with Duolingo, is to create something as effective at teaching as a one-on-one human tutor, yet as entertaining as a mobile game. That's our goal. You know, we keep at it. I mean, if you look at the past quarter, a few of the things we've been working on — very soon you'll see really cool Avatar costumes coming directly from mobile games. I think users are gonna love that. We are, we're making a number of changes regarding how we present rewards to users. I mean, you know, for instance, we're showing them as cards now, which feels really collectible; we're doing things like that. Another item that isn't exactly from mobile games, but that matters, and it also matters to mention for the free tier — one of the things we want to work on is monetization tactics that don't conflict with the free tier, and we've been finding some really good ones this quarter. One that works well is longer free trials. You know, historically, Duolingo has offered a seven-day free trial, and that has served us well. We're finding that offering longer free trials is really good in that not only does it give us more bookings, which is good, but also it benefits the user; it's not, you know, they feel good; they're like, "Ooh, I have a one-month free," for example. That's another thing we're discovering, and it's something we feel quite good about.
Thanks.
Much appreciated.
Hello there, Andrew.
Taking the questions. I'd love to discuss growth and top-of-funnel more broadly. Could you help us grasp the deceleration there? I understand the comp and everything from last year, but how should we think about what has caused that deceleration, and does it need to speed up to support DAU growth?
In terms of growth, the primary thing we focus on here is Daily Active Users growth. That's the main thing we've worked on for the longest time. As for monthly active user growth, there simply isn't a team looking at that, you know, and we do report it, but no team is looking at that. Tied to MAU growth is top-of-funnel, and that we do work on. You know, the reality is that top-of-funnel has been roughly flat for, you know, certainly this quarter, and we'd like to accelerate it; we are, you know, working on that. There's a lot with marketing that I think will be really good, particularly in under-penetrated regions. I think that's one thing. The other is just, you know, making changes to the product so it teaches better and is better for free users, which should speed up word of mouth. Again, the main thing responsible for our top-of-funnel historically has been word of mouth, and word of mouth is this interesting thing that is, you know, beautiful because it's free, but we don't have that much control over it in terms of being able to measure it the same way we can measure retention. We're doing things we think will be really good for word of mouth, but we don't have the granularity of control we have for things like retention.
Luis, I've got a question about strategy when it comes to holding onto users on the platform, correct? Your team has consistently prioritized fun.
Mm-hmm.
It feels like there's a shift happening as we move toward an experience that's more voice-forward. Could you talk about how you preserve the entertainment factor, and what needs to change as you move in that direction? Thanks a lot, guys.
I'm not sure why you'd describe it as a change, since from where we sit, nothing has shifted when it comes to fun. The thing is, we firmly believe that the toughest part of learning on your own is keeping yourself motivated. That's genuinely the magic ingredient behind Duolingo, and what's brought us this far is knowing we must keep our users motivated to learn. There's a huge gap between what people claim and what they actually do. Someone might say they'd like to pick up something new, but in the end, they'll gravitate toward whatever is most entertaining. We put a great deal of work into making it enjoyable. In our view, giving voice and speaking a bigger role in the app doesn't take away from the fun, and as a matter of fact, our metrics indicate that it doesn't. It can be quite an enjoyable thing to do, and we'll keep at it. Now, something worth grasping about fun: everything displayed on the screen — our lovely animations, and so on — it simply comes down to the fact that people are highly visual beings. You'll keep seeing plenty of those gorgeous animations and more game-style elements, even with voice present. You'll still get all the graphics and everything else, so that Duolingo stays as entertaining as we can make it. Yeah, the teams focused on making the app more enjoyable are truly going full throttle. You'll see quite a bit, actually, within the coming days — for instance, there's this fantastic feature you'll notice, which is avatar costumes. I find it really entertaining. My personal favorite: I put one on, and if you check my Duolingo account — I'm not sure whether your app shows it yet — but I'm dressed as a hot dog, and I'm loving it.
Great. Thank you.
Hi, Eric.
Hi everyone, thanks for taking my questions. I have what might be a two-part question, if that works. Luis, first, what have been the biggest lessons so far from rolling out AI — both in terms of how users experience the platform and how much content has grown over the last few months? Gillian, as AI scales on both of those fronts, how should we view the impact on margins over the longer term? Thank you both.
Great question. We're very excited about AI. So at the broadest level, here's our goal: build something that instructs as effectively as a personal human tutor while being as entertaining as a game on your phone. When it comes to the instruction piece, AI is the vehicle that gets us there. Take our Video Call feature, for example — the one designed for conversation practice. As we've mentioned, it has improved a great deal over the past twelve months. Dialogue now flows much more smoothly, and because of that, the typical user is producing roughly double the word count compared to one year ago. That's a substantial gain. And in a similar vein, as you pointed out, there's content. Looking at the volume of learning materials we released during the most recent quarter, it basically made all of our prior output look tiny by comparison. We shipped 20,500 content units — and that happened within a single quarter. That figure is roughly equal to what we managed across all of last year. And keep in mind, we were already applying AI back then too; it's simply that our proficiency with it keeps improving. Another point worth mentioning — there's a lot on our plate right now. Take, for instance, the work on models that decide which exercise should be assigned to you. From the beginning, we've relied on a system that decides which exercise goes to every individual user. Right now, we're pushing toward far deeper personalization, since that's precisely what a human tutor working one-on-one provides. Such a tutor tailors things extremely closely to you. We're quite enthusiastic about this.
When you look at it from the cost side, I'd say there are a couple of factors at play. First, there's AI adoption, which matters both for what customers see and internally — and you can see in our gross margin guidance that we expect to land around 69% in Q4, which bakes in a much heavier dose of that element in what we offer. On the opex side, honestly, we've begun noticing fairly significant upticks in what AI runs us internally, and our guidance accounts for that. We're ramping up across the board, but there's a second dynamic here too. Efficiency surges always follow AI, so costs might spike, then the team finds ways to streamline, and you keep moving ahead. Take Q1 gross margin, for instance — it came in even stronger than we'd anticipated and quite solid versus a year ago, and yet our product still contains plenty of fresh AI content, which is possible because, per unit, expenses have dropped substantially; it arrives in these kinds of waves. Expenses fall, we take on more, and we keep it in check. So when you consider margins overall, my expectation is that we sit in that 69% zone for gross margin, and we'll handle operating expenses to match.
Great, thanks. Hey, that's wonderful. Appreciate you fielding our questions. Luis, circling back to the prior earnings call too, you know, it was clearly really encouraging to see voice woven throughout the whole ecosystem. Could you talk about how that affects Max as a whole? Like, are you noticing Max members stepping down to Super? And how should we frame the product market — the go-to-market strategy for Max — now that AI is getting, not commoditized exactly, but increasingly accessible throughout the wider ecosystem?
Yeah, I appreciate that, since it gives us a chance to make things clearer. What we stated previously is that we intended to bring Video Call into Super Duolingo, our mid-tier offering. Doing so matters to us because Video Call is an excellent teaching feature, and we simply want far more people to be able to use it, since when they can, their learning improves, they'll spread the word to friends, and so on. We genuinely think this is the correct move; we've begun implementing it, so right now we're running several tests that give Video Call to Super subscribers, especially new users. Those newly joining Super are receiving Video Call. We haven't rolled this out to our entire existing user base, and so forth. Right now, nothing has changed for Max. I can't say what will happen with Max. Many options exist — it's still, again, only been two months since our last conversation, and we simply haven't completed all the various tests. There are a few ways it could go. Perhaps we reduce Max's price. Perhaps, you know, we do something where Max subscribers get unlimited Video Call while Super subscribers don't get unlimited. Several things are possible; at this point, regarding metrics, we aren't seeing much of a difference, aside from the fact that new users — one cohort of them right now — aren't even being shown Max. They're only being shown Super. That's merely one among the many tests we're conducting.
Understood. Thanks. That's useful. For Gillian, taking a broad look at the guidance here, I mean, clearly we get that bookings face a harder comparison as we move into 2Q. Could you walk me through the pluses and minuses behind that re-acceleration in the back half? I recall you also said something about 20% DAU growth having ebbs and flows across each quarter. So would that back half guidance, intuitively, suggest DAUs get better from early gains coming out of these product initiatives we've got underway? Thank you.
Yeah. Looking at the second half guidance, broadly speaking, we're simply running the business on that 20% growth basis. You might see a few early returns from the investments we're making, but I wouldn't count on much of that. Our approach this year is really to take the long view, and we want to give ourselves room to operate inside that bookings guidance range we provided, so that we can make every investment we want to make and do what we believe is right when it comes to the customers. On the growth rate, Q2 into Q3, Q2 carries a really difficult comp thanks to the launch of Energy specifically, plus a handful of other features that made bookings so strong a year ago. You'll see us recover from that comp, and then those DAU figures will keep fueling the bookings.
Excellent. Thank you to you both.
Hello. Hey, everybody. Appreciate you taking my question. The speed at which your content generation capability is growing is genuinely fascinating. On one hand, given that you now possess a truly comprehensive core offering spanning language learning, at least for the most widely spoken tongues, are you beginning to spot a chance to run A/B tests on fresh content against various engagement metrics? In the past, have you observed a rise in retention, payer rates, or whatever else, tied to better-quality content? And looking ahead, how does this shape your thinking about moving into other subjects past language learning, now that the barrier to entry has dropped?
Thank you, Nathan, for asking. Yes, something thrilling here is that, honestly, we've spent years striving to get the top nine languages covered all the way up to Duolingo Score 129, and that's now achieved. Within the company, I kept saying, "Oh, we're nearly there. We're nearly there." Someone told me something that really stayed with me: "No, no, that's only the start." Once all that content is in place, and it now is, we're genuinely positioned far better to take this material and make it, you know, markedly improve according to user performance. We've begun doing exactly that, and indeed, we've witnessed shifts from content changes—certainly in content quality, but also in which kinds of content get shown—that produce intriguing effects on retention, especially among new users. We're running many experiments to determine, for instance, that what we teach in the very first unit carries huge weight. Do we teach greetings? That's one option. Do we teach only the terms for mom and dad? That's another. It genuinely isn't as straightforward as always being better to teach greetings. It's not that simple. There's more nuance to it, yet broadly speaking, these choices do influence retention. That excites us greatly. Also, we're probably heading toward a setup where, with all this content available—which is fantastic—we might even begin producing content tailored specifically to you, drawing on everything we know about you. We might generate content just for you, perhaps not the very next exercise, but, say, the one two exercises ahead, based on all of it. That line, or that bit of material, was produced only for you; we truly are arriving at that stage, which is quite thrilling; regarding other topics, every topic comes with its own distinct hurdles and distinct aspects; certainly, AI assists us in bringing on new topics more quickly, and chess serves as an excellent illustration. To put it plainly, we managed to introduce chess in roughly 9 months. Every topic involves its own kinds of material that must be included; say, incorporating math was fairly straightforward if your goal is to include math the way, say, ChatGPT would present it, meaning a giant block of words. Should you wish to include math featuring diagrams and, as you know, interactive elements, and so forth, that becomes tougher; AI is rendering it, well, it is unquestionably rendering it simpler, yet it remains considerably tougher. You aren't merely, say, asking an AI, "Simply hand me some material." Doing it with all the diagrams is considerably tougher. I suppose my response to that is, indeed, it is rendering the addition of more topics simpler. No, it is not yet effortless to introduce new topics. Right now, we are fairly content with the topics we possess. In particular, we are extremely thrilled about math. You noticed we opened this call with a video for math. We at last reached the stage where our math course truly contains all, nearly all the material spanning grades two through 12, and it can genuinely clarify things when you got things incorrect; we are extremely thrilled about that.
Great. That's quite useful. Now, considering the balance sheet, there's a substantial amount of cash sitting there, and you know, strong free cash flow. So I suppose, what do you think is the appropriate amount for buybacks, or what might be some possible ways to deploy that cash down the road?
When we examine the cash, you noticed that we gave back a certain amount of cash to shareholders through a buyback during the quarter, and we have a $400 million authorization in place, and we're ready to use those funds. I believe that overall, our main focus is on running the business, we're putting money into the business too as we consider the business this year. It's really going to be a mix of both. Naturally, with a buyback, you purchase more when the stock price is down and less when it's up. We'll evaluate that and consider where the stock stands. We feel it's an excellent moment to purchase our stock. It's an excellent method for us to counterbalance dilution from the last couple years as we look at the business. We enjoy doing this, and as you can see from our free cash flow projections, we're going to produce nearly as much cash as that buyback amounts to anyway during the year. That's the equilibrium there. From, I think your other aspect of capital allocation is naturally M&A. We're present in the market constantly reviewing opportunities, but as you've observed, much of what we've completed is relatively small in scale; it's not going to really impact the balance sheet so severely, or if you will, in terms of, like, a major deal. Clearly, we'll always consider everything that's available, but as you've observed, Duolingo's very concentrated on expanding Duolingo, putting resources into Duolingo, and moving forward from there.
Thanks.
GameStop has an interest in acquiring eBay. That's something we might consider as well. Just joking.
Regarding that final remark, I'll let Luis handle it somewhat.
Just so you're aware, eBay is not something we're acquiring.
Could we perhaps discuss — clearly you launched and had that late-April announcement regarding advanced content now being available across every top subject. So from a marketing standpoint, can you describe how significant a unlock that represents, in terms of how you're either allocating that added performance marketing spend, now that all the content is there, and how we ought to begin thinking about how that might help refill the top of the funnel as we move through the back half of this year and into the next one?
Yeah. Regarding, definitely regarding performance marketing, I mean, this is important, I believe, most for English learners, sorry; when it comes to finding users who will join the platform, English learners are the group most interested in content that's more advanced. Something interesting about Asia is that in several big Asian markets, profitable performance marketing is something we can do. You know, performance marketing at Duolingo has been this curious situation where, since our free version is so strong, doing profitable performance marketing hasn't been simple for us, because what occurs is we bring people in, and then as free users they're extremely happy. We're discovering that we can achieve that, especially, for instance, in China, we're able to acquire profitably. And these are English learners. In that regard, these things are somewhat connected. What I'd say is the key point with performance marketing for us is that historically we've simply under-invested in it. We're becoming far more professional about it, certainly this year, and I believe you'll see those results over the next few months; we're quite excited about that, simply because we're at last, you know, constructing the infrastructure to have proper attribution to direct users to the correct place once you've acquired them, etc., something a company our size probably should have built years ago, but we kind of just overlooked it. We're quite excited about that.
Great. And then, as a follow-up, while you're running tests on Video Call inside Super Duolingo for a certain group — that group being net new paid subscribers on Super — could you share what you've observed so far regarding price sensitivity and how much extra people might be willing to pay for that feature at the Super tier? And then, Gillian, how does that shape the way you think about the gross margin profile going forward?
The key point I'd make is that we're conducting tests to determine what the correct price ought to be for Super with Video Call. I'm not able to share all the findings, since we kicked off this effort a couple of months back and it requires time. You need to set up the A/B test, then operate that A/B test over a few weeks to obtain the findings, and so on; it's only been eight weeks since this effort began. I simply don't have many findings yet. What I can share is that customers are prepared to spend more for Super with Video Call. That much we know; that's what we've discovered. The exact amount — I should have more to share in a quarter or two.
Ryan, one thing we've genuinely been pushing everyone toward is that we run the business inside a defined band of financial outcomes, and the point of that is to give us room for exactly this sort of effort, this sort of experimentation. Trying out various tactics for how we bring price to the customer. Every bit of that is baked into the guidance here, within those bands we intend to run in during 2026; in any single quarter it could come in somewhat higher, somewhat lower. We saw that coming as we entered the year and we're carrying it out accordingly. Nothing in there catches us off guard, and our view is that the numbers we've set forth for 2026 are able to absorb it.
Awesome. Appreciate it. Over.
Ralph.
Luis, hi Gillian. I hope this makes sense. I'll give it a shot. Luis, if you look back over time, there's a bit of a contradiction here. In the past, you charged too much. Perhaps right now we're in a stretch where the user base isn't being charged enough. So maybe taking a step back, could you tell us about the indicators you're watching at present, and what they tell you that you're headed the right way? And more broadly, at what point would it be appropriate to resume charging more? I realize only a couple of months have passed since the previous call, but I'd really like to hear your thoughts on this. Thanks.
Yes. That's an excellent question, since it lets me highlight a point I've made previously. You said contradiction, and that's essentially how we feel. At once we charge too little and too much; it's strange. The truth is that about 12% of the people who use our app monthly pay for a subscription; in our view that figure ought to be considerably larger. If you consider comparable freemium businesses, theirs are far above ours. Spotify, for instance, sits near 50%; we genuinely believe there's plenty of opportunity; we ought to be able to bring in additional subscribers. Meanwhile, I believe that with certain forms of monetization we likely went too far, meaning we likely introduced excessive friction into what free users experience. Essentially, the core issue was that the majority of our revenue-generating methods conflicted with growth in DAUs. A lot of them conflicted with DAU growth, since if free users face greater friction, the result is that a few more sign up for subscriptions; that's positive. It brings in revenue. But it also causes a few more to quit. What we must accomplish, and what we're currently doing, which excites me greatly, is discovering approaches to earn revenue that don't set DAU growth against monetization. Such approaches are out there, and we're quite pleased with them. I already brought up one: extended complimentary trials. At Duolingo, historically, we haven't done much testing with this, I mean how long our free trial lasts, yet when you examine other subscription companies at scale, their free trials differ quite a bit, typically far longer than ours. You notice one month. Occasionally you'll even notice three-month free trials. You're going to see us test that, and we're certainly observing — at least with the one-month test we already observe — that it raises revenue, which is great, while also not conflicting with daily active users. So, when you tell someone, "Forget a seven-day free trial — you're getting a one-month free trial," nobody's going to walk away because of that. Their reaction is, "Hey, that sounds pretty nice." What we're doing, you know, is figuring out how to monetize without hurting DAU growth. Those methods do exist; they're simply not as fast as, you know, basically putting more friction into the free experience — that's the stuff we're testing this year. You know, Gillian nailed it. The concepts we've got for this year to test this out — we'll likely try a three-month free trial. And by the way, a three-month free trial is something we've mentioned previously. Well, that's something we could never have pulled off without a year like this one, because with a three-month free trial, your bookings end up pushed back an entire quarter. You're basically showing up unable to say, "We're gonna run things under these guidelines," which is exactly what we did here. Just going, "Oh, sorry, every bit of our bookings will land next quarter. They didn't show up this quarter" — that was a little strange. That's the reason we're running things the way we are.
All right, excellent. Thank you, Luis. I appreciate that.
All right, thanks. I would like to — well, I want to inquire about gross margins. It appears your guidance suggests that they will sort of step down, or something like that, to the high 60s during the fourth quarter. Is there any reason to believe those margins hold, gross margins hold at that particular level? Is there any reason to believe they ought to bounce back up higher, or go lower? Just discuss how gross margins will move going forward, like how should we think about what pushes them up or down after this year.
Mark, whenever we consider any of the margins that include AI content in them, so let's, you know, take gross margin, what you usually discover is that when you begin to roll out features, they can be more costly for us, and afterward we refine that expense over time. If you look at Q1, for instance, the margin stayed really strong relative to the prior year, and that happened because our AI costs per unit have dropped significantly. Looking ahead, we truly want to include more and more AI as an ingredient in the product, which is why we have the margin guidance that declines to 69%, which is basically where we stood last quarter, as well. That simply reflects a lot more AI content, which we believe is wonderful for the business over the long term. You know, I think from a positioning standpoint, it's possible we might refine that further, but we kind of want to be placing that much AI into the product. I think 69% is a solid level to think of us leaving the year at. If that pans out, that's going to give us nice growth too. Having said that, this is an evolving environment and, you know, certain optimizations arrive sooner than you'd anticipate, and you might see, you know, movement in either direction on that one.
Okay. Thanks, Gillian.
Justin, hello.
Thanks so much, and good evening. You know, when I think about Duolingo, I see a company that's always had a product philosophy built on moving fast with A/B tests, and coding has definitely made it simpler than it's ever been to run plenty of those. I'd really like to hear how you're viewing engineer productivity overall, how many tests are being run, and how we ought to think about that possibly shaping headcount needs down the road. Thank you.
Yeah, that's a wonderful question. You're correct. We do a lot of A/B testing. At any given moment, hundreds of A/B tests are running simultaneously. That's been our product philosophy, and that's how Duolingo has improved over the years, simply because we can keep raising things. The moment we focus our attention on a particular metric, give us just a couple of months, and normally we can raise that metric. We simply conduct a large number of A/B tests. What we're seeing is that the count of A/B tests we're able to conduct keeps growing. Our belief is that this comes from AI usage, especially within our engineering and product organization. The growth isn't enormous, but it's sort of the first time in years that we've witnessed growth on a per capita basis; we do believe that's contributing. I'm not sure where this will ultimately lead, but it's a fascinating thing. Consider last year, for instance — if you'd gone back a year and read Twitter, you'd have been led to believe that you can program anything you like in five seconds and it's finished, and why wouldn't you be running 10,000 A/B tests simultaneously with just one engineer? That's an exaggeration. That simply isn't true. Until quite recently, we and companies at our scale hadn't observed a genuine rise in velocity, broadly speaking, but we're beginning to observe that rise. It's still modest, yet we're beginning to observe that rise, so I'd assume you'll keep observing that rise. I doubt we'll manage to run 10 times as many A/B tests per capita, per engineer, but, you know, it is growing and that's something we're quite pleased about.
Thank you.
Hi, John.
John.
Hey, great. Thanks for taking my questions. I just wanted to ask about DAU trends, sort of. Could you give us some color on how U.S. DAUs are trending compared to international DAUs, and what that relative geographic growth might mean for bookings over time, since U.S. users generally adopt a subscription at a higher rate than international users?
In the U.S., DAUs are growing, and they're growing in pretty much every country. It is true that they're growing less in the U.S. than in many international markets; in particular, Asia is the fastest-growing market that we have, that whole region. When it comes to the impact on monetization, that really doesn't move the needle much, as far as I can tell. It's simply that the U.S. does monetize well, but it also happens that plenty of other nations monetize fairly well too. China is a great illustration. China's monetization is roughly on par with Western Europe — so roughly comparable to France — which isn't as high as the U.S., but it's still fairly high. Considering China's growth rate — I'm not sure what the most recent figures are — but it's quite substantial, I'd say. I simply don't believe that slow U.S. growth automatically implies our bookings growth must be really low. When it comes to the U.S., my expectation is that improving how well the product instructs people, along with stronger word of mouth, will help. We're also putting some money into marketing there, something we hadn't done before in terms of paid efforts. My hope is that together these moves will lift our U.S. year-over-year growth above where it stands today.
Okay. Thank you.
Hi, Shweta.
Hi there. Thanks for letting me ask. AI-powered content generation led to a significant boost in how quickly content gets produced. So I'm wondering if you could explain what you're doing to keep content quality under control as both the amount and engagement keep climbing. Thanks.
Yes. I'd say we put in a great deal. The biggest factor behind our content not expanding at an even faster rate is that we're determined to guarantee it stays extremely high in quality. There are several measures we take. For one thing, we run assessments of our content, using both AI and people, to help confirm that it's of very high quality. Once that's done, we try out the content with a limited number of our own users to check whether it's high quality, and when it is, we then expand how many people receive it. We're working hard to be extremely cautious about ensuring the content is high quality. That's a fair observation, since the truth is that no matter how impressive AI may be, without care about quality you can end up with a great deal of slop; we're putting in a lot of effort to prevent that from happening. Our content quality, I believe, has in fact gone up over the past two quarters. How we can tell is that we essentially run spot checks, and we sort of score the quality of everything, and we've found that across the last two quarters, quality has genuinely risen.
Okay. Thanks, Luis.
Yeah.
Hey there, Omar.
Hello, and thank you. How are things with you? Okay, I'd like to circle back to performance marketing briefly. It's good to hear the company now takes that area far more seriously. When we last talked, my takeaway from what you said was that down the road this product would be dramatically superior to what exists now, and that it would genuinely transform how language learning works for people. So when it comes to the product's own maturity, does that hold back how much we can scale spending on performance advertising? The reason I ask is that performance advertising usually aims to optimize particular kinds of user behaviors. Is that what's happening? And second, do you have a sense of when you'd be prepared to truly push hard, assuming your team has finished all its testing — when would the product be set to go all out with performance marketing?
Yes. What I'd say is that for us, the thing limiting performance marketing has been, number one, creating the infrastructure so it can function as a far more robust performance marketing operation. We're working on that right now. That's one piece, but there's a second piece, and it isn't about teaching quality — it's mostly an issue with how strong our free offering is. Depending on the region and depending on what we're advertising, one of our difficulties has been bringing in a user, having them arrive, and then actually converting them to a subscriber, rather than bringing in a user, having them arrive, and finding they're perfectly content users of what we give away for free. That has been the key constraint, and it's what we have to move past. Right now, in certain geographies, performance marketing is profitable for us, though in a lot of geographies it isn't.
There's just one more point, I think.
All right.
What I meant was, that's the sole item I'd tack on.
Go ahead.
...what's happening is we are putting money into marketing during this year, and that isn't only on the performance side. Correct? Our team takes a layered approach when it comes to marketing, and to increasing that spend, and it's genuinely carefully considered and has a fair amount of variety to it too.
Quite a bit of variety there.
I just want to verify I'm not viewing performance marketing incorrectly, since I figured it would be hard to run performance marketing for something that isn't steady, or for something that isn't, like, grown-up and settled, because, you know, with users, you don't really know what it is you're promoting, you know, when it shifts that much. I simply want to confirm I'm not looking at it the wrong way.
That's not how I'd put it. I mean, you know, Duolingo has existed for 15 years; it has never remained the same, and it never will. I don't think that's, you know, well, that isn't going to change, but I don't think that's been the issue.
Appreciate it.
Right. Hello, I really appreciate you taking the question.
Hello there.
Thanks for taking it. Hi there. You brought up how quickly China is expanding. Over the past 12 months, I believe there have been 2 tie-in deals with brands that worked out really well — one being Luckin Coffee a year ago, and I think another being Meituan in March. Is there anything we can learn from how well those brand tie-ins have performed in China, and any way to carry that success into other markets during the coming year? Thanks.
Thanks. Well, you picked up on those brand partnerships. In China, we've enjoyed amazing brand partnerships. Part of that, I think, is simply that our IP and our brand are extremely strong there, and because of that, some of the biggest brands want to work with us. To give you an example, we'll have a partnership with McDonald's in China very soon. The really big brands that want to work with us, they approach us about doing it. That's one piece of it. Another piece, I think, is that brands over in China — brands such as Luckin Coffee, say — are simply far more willing to enter partnerships than, say, brands in the Western world are. For instance, you just don't see Starbucks swapping out everything in all its stores every 2 weeks for a new brand, while that's essentially what Luckin does. So there's something to learn there. For sure, that team is — well, our partnerships group in China and our marketing group in China are, you know, performing at a very high level. There are things we can learn that we could apply elsewhere, especially in Asia. I'd add that part of this also comes down to the fact that it's the China market. And one more thing I'd mention: China isn't expanding quickly only thanks to those great partnerships. I mean, the great — actually I think it's the reverse. I think those great partnerships are arriving partly because we're expanding quickly and we're viewed as a really cool brand. The truth in China is that there is simply an enormous appetite for learning English that just keeps expanding and expanding, and that's the primary reason China's is expanding.
Thanks so much.
Hi, Alec.
Hi, Luis. Hi, Gillian. Appreciate it. Regarding how DAUs connect to top-of-funnel growth, compared with the sort of visibility you've described concerning the shape of this year's bookings inflection, which early tests — or perhaps tier or geo mix — exist that give you that visibility about the inflection?
Sorry, regarding, an inflection of.
Regarding bookings as we leave the year.
Yeah, so starting with bookings, I think when you examine how things move quarter by quarter, the guidance we're giving as we move from Q2 to Q3 to Q4 is basically in line with where this company has stood over the past couple of years. One point we want to be sure everyone keeps in mind is that we're in this for the long haul, and the investments we're making are. We might witness some results from them during this year, but our focus really extends past this year. When it comes to 2026, it's much more about running at that 20% DAU growth and then expanding the business. At the moment, what you're seeing, if you look at that progression, is pretty normal seasonality. There's been an adjustment in Q1 and Q2 tied to our new monetization balance, but after that, in Q3 and Q4, what you're seeing is quite standard for us.
OK, that's great, thanks.
Appreciate it. Thanks to our operator. I want to say thanks to all of you for being with us, and we're eager to connect with you again on the following call.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Duolingo, Inc. transcript - plus 132,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Duolingo, Inc. earnings transcripts and 132,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.