Home / Transcripts / Duolingo, Inc. (DUOL) · November 5, 2025

Duolingo, Inc. (DUOL) Earnings Call Transcript

November 5, 2025

NASDAQ US Consumer Discretionary Diversified Consumer Services earnings 52 min

Earnings Call Speaker Segments

Luis von Ahn executive
#1

Hello, everyone, and thank you for being with us today. Our third quarter was robust, delivering strong results on every measure, and we remain on course for yet another outstanding year. Over 50 million individuals now engage with Duolingo daily, and our guidance points to close to $1.2 billion in bookings for this year, reflecting 33% growth alongside an adjusted EBITDA margin of 29%. Taken together, this shows we've quickly grown our impact while also increasing profitability, and even so, we sense we're still at an early stage; we think AI is set to radically reshape education, and we can see a clear path toward creating a product that instructs better than at any point previously. That is why this is such a thrilling time for Duolingo, and we'll now begin taking your questions.

Bryan Smilek analyst
#2

Thanks for taking my questions. Luis, to begin with, could you give us a clearer picture of what's actually fueling the DAU increase as we head into Q4, along with engagement more broadly? And how near do you believe you are to being back on track with U.S. marketing? You brought up rising impression volume, but has that begun showing up as user growth yet, or what's the right way to think about it for Q4?

Luis von Ahn executive
#3

Yeah, that's a great question, Bryan, thank you. We just reported 36% year-over-year growth for Q3, and we feel good about that. There are pluses and minuses in terms of how we arrived there. Looking at what went well, the Luckin partnership we mentioned in the shareholder letter is a good example — it drove a meaningful increase in users across Asia. Retention also benefited from several product enhancements, which was encouraging. Conversely, as we noted previously, we temporarily stopped publishing unhinged content on social media while we listened to our community and worked on building brand love. When we aren't putting out unhinged posts, the odds of anything going viral drop substantially, and as a result, DAU growth took a hit. Here's the brighter side: during the past several weeks, we resumed making those wild posts on our social media accounts. It hasn't yet returned to the highest point it once reached, but we've seen substantial recovery. That's now becoming visible, and we anticipate it will have a favorable impact on DAUs. Regarding DAUs in Q4, we're expecting a certain degree of slowing down compared with Q3, though we feel quite good about the level at which it has settled. I should note that we aren't providing Q4 DAU guidance, but I can tell you that September and October each saw DAU growth of roughly 30% year over year, and that comes against a fairly tough comparison from last year's quarter.

Bryan Smilek analyst
#4

Excellent, that's really useful. And Luis, I believe you brought up three main pillars as well: making money, expanding the user base, and simply how well the teaching works. Could you expand on what's actually behind the choice to move spending toward the long-term view? Is it the chance that AI presents, or how should we consider the effect this has on bookings growth? Obviously you spoke about 25% plus being your North Star a while back, so I'm just wondering how we eventually return to that level.

Luis von Ahn executive
#5

Yeah, well, as you said, and as I wrote in the shareholder letter, there's an enormous opportunity at this moment; we're seeing an enormous opportunity. In the coming years, education and the manner in which people learn are going to transform at a fundamental level, and that's due to AI, and AI also gives us a clear path now to build an app capable of teaching extremely, extremely effectively, far beyond anything humanity has ever witnessed, on par with a human tutor, yet at the same time more engaging, and if we can pull that off — at present we have, I'm not sure, we just reported 135 million monthly active users. If we can make an app that teaches at that level, vastly superior to what we offer today, we'd be discussing billions of users, and that's what we're aiming for here; so that's the reason we're putting money in for the long haul; and what it involves is directing a greater share of investment into areas such as better teaching, which, when we teach better, ends up supporting user growth, though there's a delay. Each time you enhance your courses, users do increase, but it takes time for that to show up. Then user growth has a delay before it reaches monetization since people need some time before they subscribe; so this is essentially a long-term matter, but we're extremely optimistic about it, and that's the reason we're taking this approach; and the aim here, given how big the opportunity is, the aim here is to grow DAUs quickly for a very extended period.

Bryan Smilek analyst
#6

Thanks, Luis and Matt, appreciate it.

Nathan Feather analyst
#7

Hey, everyone, and thanks for taking my question. You're describing a really large opportunity in front of us. So is this emphasis on prioritizing for the long term a lasting change to the strategy that will keep going, or do you intend to eventually go back to the prioritization mix we have now? And related to that emphasis, given how durable this shift is, do you think the bookings pressure you're seeing in Q4 will carry over into next year? Thanks.

Luis von Ahn executive
#8

All right, reaching the stage where our app is, in my view, the absolute finest method for mastering any significant subject — that's still several years off. So expect us to keep putting money in for quite some time, and that's a key point to understand. That said, I want to be clear we aren't actually providing guidance toward, say, the coming year, yet plenty of the product efforts we'll launch during that coming year have us really enthusiastic. Our video calls aimed at newcomers are set to improve dramatically — we're branding this Guided Video Calls. The app will also become far more social; we're going to include all. Right now, the Math course only has a little bit of content; in the next few months, we're going to have all of the Common Core K through 12 content in the math. Our Chess course is going to have player versus player, and it's doing super well and growing really fast; we're going to have a full revamp of our Music course. And for the top nine languages that we teach, we're going to be able to teach from zero to Duolingo Score 130, which is where you can get a job in that language. There are simply many developments we're really enthusiastic about. And Matt, I'm not sure whether you'd like to say anything further on that.

Matthew Skaruppa executive
#9

Yeah, Nathan, I'd like to frame this within what we've previously discussed. So for quite a while, we've been telling all of you and our investors that our goal is to maximize platform LTV. What Luis is talking about is a minor adjustment at this point that will assist us in growing users over a long period, getting users to complete more lessons, learn more effectively, devote more time to the app, and generally become more engaged Duolingo users. And we believe that if we accomplish that and accomplish it well, we'll both grow users over a long period and raise platform LTV over a long period. So I think that will. Ultimately, we come down to bookings and the financials too. I believe we provided guidance for a really strong 2025. We think this can coexist with strong financial performance, and this is simply a balancing act that we've always performed, and we're informing all of you that we want to ensure that right now, we're balancing it so that we can grow very quickly over a long period in the future.

Nathan Feather analyst
#10

That's really useful, thanks.

Wyatt Swanson analyst
#11

Hi everyone, thanks for taking the question. I'd really like to learn more about where things stand with the new chess course. At Duocon, I believe you said there are millions of daily active users. On that note, are you noticing any variation in engagement or retention among chess course users compared with those in your main language courses? And related to that, I'm also curious about your new PVP offering. Could you share when you anticipate it will be completely rolled out?

Luis von Ahn executive
#12

Yes, so obviously we're extremely enthusiastic about Chess. Among all our courses, it's the one expanding most quickly. Its growth outpaces Math and Music by a wide margin, and it's also faster than the pace at which languages originally grew. That's accurate. We have millions of users. We won't state the precise figure, but it has already moved ahead of Math and Music. Where retention among Chess users is concerned, it hasn't existed for very long. Our Chess course has been available for three, four months, yet based on what we've been able to track across those last three to four months, it's marginally above language learning. So that makes us quite pleased. Also, as you brought up, we began rolling out PVP. That stands for player versus player, meaning individuals can play against one another. Right now, 50% of users on iPhones, on iOS are able to see it. It isn't on Android yet, though it will arrive on Android fairly soon. So we anticipate that over the coming few weeks and a small number of months, everyone who has the Duolingo app will be able to play PVP chess. And across the next year, we're expecting considerable growth from chess, and I'm very pleased with it.

Wyatt Swanson analyst
#13

Thanks, that's helpful. I've got one brief additional question. When you say the focus shifts toward expanding the user base rather than driving revenue, what does that actually mean, and how will the app itself be different?

Luis von Ahn executive
#14

Sure, let me walk you through an illustration. The thing is, every time we test something, we've had to weigh competing considerations. Certain tests boost every metric. That's simple — you just ship it, since all the metrics go up, and that does happen sometimes. Other times, though, a test lifts one metric while dragging down another. Here's a made-up scenario. Suppose free users currently receive 25 energy units when the day starts, and each exercise they complete uses up one unit. Imagine we tested lowering that number from 25 down to, let's say, 24 — meaning one less energy unit each day. We're aware that would bring in more revenue. That's simply how it works, since more users would exhaust their energy, leading more of them to want to pay for a subscription. But at the same time, we're aware it would lower daily active users, because it would annoy some portion of them. We've always faced these kinds of judgment calls. What we're saying is that we're now making a change: for this kind of judgment call, user growth will take priority over monetization. So with that made-up test I described, we would not ship it — we wouldn't move from 25 to 24 energy units — even if it meant a meaningful amount of bookings gains, if it caused a genuine decline in daily active users. That's the kind of thing we're doing at present, and once more, as a reminder, we're doing it because the opportunity in front of us is so large that growing quickly over a long stretch of time is simply good for us.

Wyatt Swanson analyst
#15

There you go. Thanks, guys.

Matthew Skaruppa executive
#16

Thanks, Wyatt.

Ralph Schackart analyst
#17

Thanks. Evening, Luis and Matt. Both of you, following up on that thread of questions, perhaps the core one is: why at this moment? What indicators suggest this change, or apparent change, toward prioritizing growth rather than near-term profitability because of AI progress? What's driving it? Then could you help us understand how long this pivot or shift lasts? Will it run, say, through all of 2026? Or is it shorter in nature? Any additional detail would be welcome. Thanks.

Luis von Ahn executive
#18

Thanks. On the timing question, that's a good one. What's true is that across the past 2 years, it has simply grown more and more obvious that we sit at a singular moment, especially regarding education — both in terms of how teaching will unfold globally and how effectively we at Duolingo can instruct. We observe this directly in our own metrics, in the speed at which we ship content using features such as Video call. Each month we notice how much better it gets, so this has been building for some time, and what occurred is that during the past 1 to 2 months, I truly aligned the company behind this change. Honestly, it isn't as though a single day arrived when I got up and concluded, "Oh, let's pursue that." Rather, we genuinely united the company to say, "Look, the opportunity before us is enormous. Let's make it a priority to ensure we can expand across a long stretch of time and also build an app capable of teaching truly better than anything previously seen." Regarding the length of time required, that's a fascinating question. I believe what you're getting at is something like, "Well, will this damage bookings, and will it damage bookings indefinitely?" That isn't how I see it. Seeing financial outcomes from these long-term investments we're making will require some time. Yet we will behave for quite a while as though an immense opportunity exists, because one does, until we capture it; however, I believe we'll witness solid outcomes from this even considerably earlier than that; so we aren't saying, "Oh, discard all the bookings or anything of that sort."

Matthew Skaruppa executive
#19

Right. The one point I would put on top of that, Ralph, is this: you've watched us handle that balancing act across the last 3 years too. On average, users rose 55% each year during that stretch, while bookings were up roughly 45%, and through all of it we kept striking comparable trade-offs. What's happening now is just a modest shift in emphasis as we work through them, so it isn't as though none of these choices were being made before. You can see it reflected in the rest of the year guide along with the Q4 guide — the effect there was minor. It isn't very large. Would we anticipate part of that carrying over into 2026? Absolutely, though once more, to frame this broadly, the financial hit from this type of reprioritization is fairly small; and in our view it's justified since, as Luis put it, the opportunity is enormous. So the balance of risk and reward looks appropriate.

Ralph Schackart analyst
#20

All right, excellent. Thank you, Luis. Thank you, Matt.

Alex Sklar analyst
#21

Hello, Alex. Hey, everyone. I appreciate you fielding the question. I wanted to build on what Wyatt asked and perhaps on Ralph's comments as well. Ralph mentioned something near the end there — regarding how to frame just how significant these shifts could be, is it simply a matter of paying somewhat less attention to paid conversion so that the freemium experience improves right now? Or is there a wider set of ideas, perhaps around shifting video call down into a few of the cheaper plans? And then I have one more thing to ask.

Luis von Ahn executive
#22

All right. One way to look at this is what you brought up first; it really is that straightforward in certain of the tests, which, incidentally, not all tests. A lot of the tests we conduct merely enhance metrics. That's fine. In certain tests where judgment calls come into play, we will tilt the balance somewhat further toward user growth. This isn't an enormous shift, yet it is a shift. As for whether we will relocate video call into additional tiers, and so on, that probably will occur. At minimum, we will probably try to make that happen, though that has nothing to do with this. Regardless, we are constantly considering relocating various features across the different plans, and we will test it. It could turn out that — I'm not sure about video call, but it could turn out that certain Max features work better in Super or perhaps even in the free tier, and we will test every one of those; and as we test that, we are working to optimize lifetime value of our platform LTV; we are working to optimize for that. So you might see us test things along those lines.

Alex Sklar analyst
#23

Alex, thanks for the question.

Matthew Skaruppa executive
#24

Matthew here — I'm happy you brought up free-to-pay conversion, since that's precisely the way it actually shows up in our business. Let me make sure this is clear. Take what Luis mentioned: during the last 2 months or so, he got the whole company moving. Back in September, we observed part of this, and the pattern was a modest dip in free-to-pay conversion — yet conversion on that free-to-pay basis was still up versus the prior year, so it remained healthy. The change was only at the edge; it came in a bit lower. So when you're modeling the financial effect, that instance is a good illustration. You're correct to note that this is the mechanism through which it would come.

Alex Sklar analyst
#25

Alright, that's really useful context. Then, as a follow-up, Luis, during the previous call you mentioned this concept regarding video calls and the typical count of words spoken in each session. It was sort of a fresh metric you intended to begin evaluating. What insights have you gained thus far, now that you've been working on improving that metric? And what sort of schedule should we expect for incorporating some of those adjustments into the product? Thanks.

Luis von Ahn executive
#26

I appreciate that question. Indeed, this is genuinely an excellent metric, and we've succeeded in improving it. It's crucial to understand this about our organization: when we commit to a metric, we're extremely effective at improving it. And with this specific metric, we've been improving it. It has greater than doubled during this year when it comes to the typical count of words spoken for every Max subscriber. So we're quite pleased about that. Regarding modifications to Video Call that will arrive shortly, one I'm really enthusiastic about is that currently Video Call is a monolingual experience in whichever language you happen to be studying. So if you're studying Spanish, everything is in Spanish. That works well for practicing Spanish, yet for beginners, it's excessively difficult. If your vocabulary is only 20 words, carrying on a complete conversation solely in Spanish is extremely challenging. Therefore, what we're currently testing is something we refer to as Guided Video Calls, which are essentially bilingual. So, if you're an English speaker studying Spanish, this would be partly English, partly Spanish, and it's far simpler for beginners; we're observing that when beginners receive this, they genuinely say more words for each call since they can actually accomplish something. Consequently, we believe this will significantly boost Max conversion. Incidentally, I ought to mention a couple of points. The majority of our users, and definitely the majority of our Max subscribers, are beginners; therefore, we truly believe this will boost Max conversion. The second point I'll mention is that with these Guided Video Calls, we aren't promoting them at present, nor are we employing them to convert users into Max subscriptions at present. So we've released them, and the following phase is essentially to inform non-subscribers that these are available so we can encourage them to subscribe. Thus we're quite enthusiastic about what that might accomplish for Max.

Alex Sklar analyst
#27

All right, great. Thank you both.

Matthew Skaruppa executive
#28

Thank you.

Ygal Arounian analyst
#29

Hi there, good afternoon, everyone. Apologies for the noise in the background. Regarding AI and the idea of making education better than it's ever been, the way you describe it, does that speed up your plan for rolling out additional language learning modules? I realize that's within the ones you already offer, but also expanding into new subjects — and what is it about the current shifts around AI that makes this possible for you now?

Luis von Ahn executive
#30

Yes, the kinds of things you're going to see — it certainly speeds up our plan when it comes to broader language coverage; that isn't the same as new languages. The truth is, when it comes to languages, what people wish to learn is extremely uneven. At this point we offer instruction in 40 languages. For the remaining ones we don't cover, demand is quite small. However, the top nine languages we do offer — these are roughly the Spanish and English and French and German and Italian, the major languages people are eager to learn. Those top nine make up the overwhelming majority of our user base. And what you're going to see from us is that we'll move more quickly when it comes to putting content into those top nine languages. Currently, for the majority of them, we don't take you to the point we'd like to take you, which is a Duolingo Score of 130, equivalent to a CEFR level of B2, which is the level where you can land a knowledge job in that language. You'll see that during the coming months, we'll be putting in content capable of achieving that across all of the top nine languages. The other thing you'll notice is that we'll simply be introducing far more varied modules in how we deliver language instruction, ones that are simply much cleverer at teaching you. I mean, they'll adjust much more effectively to you. And you'll also see us rely far more on tools that leverage AI behind the scenes to enable far more free responses, so that it adapts much more to you. As for — we're also going to apply AI to other subjects. We're applying it quite extensively for math, to get a lot more content out there; so that's something we'll do; as far as introducing other subjects, right now, we aren't working on any others. I won't claim we won't introduce other subjects next year. That could happen. As you witnessed with chess, it took us nine months from the idea to actually launching. So it's possible we'll introduce other subjects next year, though a bit unlikely, but it's possible. Still, right now, we aren't working on any other subjects.

Ygal Arounian analyst
#31

All right. Then perhaps a second wide-ranging question regarding AI. Could you describe what you're observing when it comes to compute expenses, with gross margins arriving somewhat stronger than anticipated? And is this happening more quickly than you'd expected? And what kind of effect is that having? A question we frequently receive concerns the competitive environment overall and the way AI is changing it. What's your read on that? Thank you.

Luis von Ahn executive
#32

All right. When it comes to cost, well, costs are declining. They've dropped even though we haven't lifted a finger; expenses keep falling. For our company, optimizing costs hasn't been the number one priority. Right now, our number one priority is simply creating the finest experience possible for our users, one that teaches best and is the most engaging. Occasionally, when we spot easy wins on the cost-optimization side, we'll take them; but we don't have everyone on staff working to trim costs. The reason we're able to take that approach is that the majority of our AI features, or at least the ones with the highest expense, sit behind Duolingo Max. And since its price is sufficiently high, using AI is profitable for us regardless; that's why we aren't frantically trying to cut that cost; regarding the competitive environment, I think what people tend to say—the two things they say about competition with AI—are, first, why would anyone bother learning a language through Duolingo when ChatGPT can just teach it to you? All right. That doesn't worry us much. We've made this point before. What we excel at, beyond teaching effectively, is keeping users engaged. To learn a language, you must stay engaged for years, truly; mastering a language takes years, showing up daily. We have to keep you engaged and actually practicing. Beyond that, we also need to offer curriculum spanning years so you can keep going. With ChatGPT, you can visit it and request a handful of words here and there, yet you can't really get curriculum lasting years that provides that instruction. So that particular concern doesn't trouble us much. Then there's the other point people raise as a worry: oh, well, no one will bother learning a language since simultaneous translation will exist. All right. That doesn't worry me either. I believe at 100% of the Google I/O conferences across the past 10 years, they've demonstrated simultaneous language translation; they do so annually. And it's impressive. It functions. Yet this has been going on for 10 years now, and the interest in learning languages hasn't dropped in the slightest. Actually, that question has been raised. The main explanation, I believe, is that our user base splits into two major groups. The first large segment consists of individuals studying a language purely for enjoyment. Whether a machine is capable of doing it doesn't really matter — incidentally, the situation with chess is quite similar. Even though computers outperform humans by a wide margin at chess, millions of people still want to study it. So when something is a pastime, that concern is irrelevant. The second major segment of learners using our platform are those studying English who genuinely have a desire to master it. From their perspective, having to pull out a phone and hold it up — that simply isn't something they're interested in doing. That's why it isn't a major concern for us; it just happens that tweeting about it is something people enjoy, but it doesn't worry us.

Ygal Arounian analyst
#33

Yeah. Great. Thank you.

Matthew Skaruppa executive
#34

A few things I'd like to mention here. Regarding the expenses tied to AI along with Max — I want to be certain that everybody understands — Max accounts for 9% of the subscribers we have now; its bookings grew by double compared with Q3 of the prior year. So in that respect, it's obviously performing nicely; even so, it's falling short of the ambitious targets we set for it; we had anticipated somewhat more. That's the reason Luis mentions guided calls along with the rest of the steps we intend to take so it can reach what we believe is possible; and lastly, since the costs associated with AI have declined, I don't want people to conclude that we've lost our appetite for investing. As Luis describes this pivotal juncture we find ourselves in, where we can pursue a really big opportunity, we've demonstrated our ability to expand to this size with remarkable profitability. For the year, we're guiding toward an Adjusted EBITDA margin of 29%, which sits extremely close to the long-term range we target for Adjusted EBITDA margin. And with that being the case, we will not hesitate to put money into innovation; so those investments will be made as time goes on.

Ygal Arounian analyst
#35

Thanks, guys.

Mark Mahaney analyst
#36

All right. There are a few items I wanted to cover. First, I'm aware that earlier this year you made some pricing moves or raised prices. How have people responded to that? Second, my understanding is that you planned to pause any additional increases — I believe this applied only to new subscribers joining the standard plan — but what are your views on introducing further price hikes? Given that you're now putting user growth first in the fourth quarter and afterward, I doubt you'll pursue that at this point, but please address it. Then third, let's confirm that the slowdown in bookings and revenue growth during Q4 stems mostly from your plan to reduce the free-to-paid conversion rate and instead concentrate on growing users. I simply want confirmation that this is the primary cause, rather than any decline in retention among paying subscribers or elevated churn among them, perhaps triggered by the higher price. Thank you.

Luis von Ahn executive
#37

Well, I can handle a few of those. I'll address the final one. Right, what's happening in Q4 is essentially the result of this pivot toward longer-term initiatives, which includes user growth and devoting some of our — not only are we putting user growth first, we're also directing somewhat more resources toward better teaching, and we're pulling some of that from what we do on monetization, so that's essentially the situation you're observing, and it isn't an enormous change, though it is a shift; regarding raising prices, you'll observe that we'll be testing prices; we'll be testing many different things, and you'll see us roll out whatever we believe benefits the entire platform. I can't predict which approach will succeed, but you'll watch us experiment with prices, and I mean both directions. We'll likely try some prices that are far lower, or a bundle, and here I'm discussing possibilities we might pursue, though I'm unsure whether we'll ultimately follow through should the results displease us, but we'll likely offer a bundle priced at roughly half, something extremely lightweight; those are the sorts of moves you might witness from us.

Matthew Skaruppa executive
#38

Right. And Mark, to kind of wrap this up, I'd say there are a couple of things. First, we've discussed raising prices as a price point quite a bit. Occasionally, that's something we do. Luis just mentioned that we'll keep testing that, and it's going to keep happening. Throughout this year, our ARPU has risen each quarter, roughly mid-single digits; the guide reflects that too. And that's come mostly from Max rather than from changes to price points. I'll note that at different moments over the recent period, we did raise pricing. And that shapes how much we're able to discount during our single discount event of the year. So once more, with a higher price, you can try out different things regarding how much discounting you do; and one thing that comes up each Q4 when we discuss it is the capability within our Q4 bookings guide, since Q4 is the quarter where we see the most variation because we run this New Year's promotion; this year, we did Energy, which is a core pricing mechanic. And never have we run a New Year's promo with Energy; and so as we move toward the end of the year, we're going to be experimenting with all sorts of things regarding the promo — how we run it, how we show it and display it, when we run it, all of these type of things. So that's also baked in here in the guide.

Mark Mahaney analyst
#39

Alright. Thanks, Matt. And thanks, Luis.

Luis von Ahn executive
#40

Thanks, Mark.

Ryan MacDonald analyst
#41

Ryan MacDonald here, thanks for taking my questions. Luis, if you don't mind, I'd like to stay on this topic with you for a moment. I want to apologize ahead of time for the question. The way people pick up a language varies across age groups and demographic segments. With advanced English learners, we've already observed that their needs from Duolingo can differ from what the traditional core base wants, I think. So could you discuss how, or whether, you plan to go after particular demographics? Gen Z appears to learn differently than Millennials do, especially when it comes to some of these product updates. And after that, what metrics should we be looking at to confirm that this approach is working? Sure, so user growth could get a boost from certain adjustments, though that doesn't necessarily guarantee the kind of engagement that sticks — the sort where monthly actives reliably turn into daily actives. I know I'm rambling again, sorry about that, but what specifically are you going after with these changes, and what's your method for gauging whether they work?

Luis von Ahn executive
#42

Right. So you brought up the point that it's accurate — different individuals do have different ways of learning, that's correct — yet it might surprise you how much more people have in common than what sets them apart. The truth is, and this isn't only about age cohorts, it also comes down to where people live. You know, there's this constant talk along the lines of, "Oh, folks over in that nation behave this way," or "people from that particular place act like that." Yet what we keep discovering, over and over, is that beyond the fact that many individuals pick things up in fairly comparable ways, the very factors that drive greater engagement with our product tend to look quite alike no matter which market you're in. Take something like a streak — it performs well everywhere, so the similarities are substantial. Which means that right now, what you'll observe from us is simply an effort to build a superior course aimed at the general population. That is where the bulk of our energy is going to go. Naturally, each learner's experience gets tailored, and the area where that tailoring matters most is likely the speed at which someone picks things up, since that varies from person to person. It's simply a fact that aging brings a slowdown — nobody disputes that. Having gotten older and slower myself, I can attest to it. But adjusting for speed is quite straightforward; we've been handling that fairly effortlessly for some time now. Turning to the measurements we'll be watching, user growth definitely stands out as a key one that has our attention at the moment. This is likely the single most critical number for our business, and we'll be watching it closely going forward. Here's the key point about getting better at instruction, and I mentioned this earlier: those gains don't show up right away as more users. When you make a course substantially stronger, over time learners may start to notice that they're absorbing the material somewhat better — that lifts retention, or maybe more people hear about it because students are recommending it, saying things like, "This really worked for me. You should try it too." So there is a knock-on effect. We're aware that better teaching carries over into growth in users, though it doesn't happen right away; that's essentially what we're referring to as long-term. What we'll be examining in that area includes simply gains in learning outcomes. We're able to measure how effectively learners are progressing. The encouraging part is that nearly every single year since we began tracking this, Duolingo has in fact been instructing more effectively than it did 12 months earlier; we'll likely observe that our teaching quality improves at a quicker pace than previously, since we're embracing it and dedicating greater effort toward it. That's probably what we'll witness. Our theory — and it is a theory I strongly have faith in — is that this will carry over into growth in users. It simply won't be linear or fast.

Ryan MacDonald analyst
#43

That tracks. Alright, I'm also getting up there in years and moving more slowly, Luis, so no issue at all.

Luis von Ahn executive
#44

Isn't that true of all of us? Isn't that true of all of us?

Matthew Skaruppa executive
#45

We all are.

Luis von Ahn executive
#46

But wiser.

Ryan MacDonald analyst
#47

Right, that's what we're pulling for, too. Matt, I get that you probably don't want to get into any kind of detailed talk about 2026 guidance on this call, but obviously there's a ton being spent on courses and on content. Can you give us a rough idea of how big those numbers might be? And do you still think it's possible to keep pushing EBITDA margins higher while all that's going on, looking out over the next two years or so?

Matthew Skaruppa executive
#48

Yeah, I hear you, Ryan, and I appreciate it. You were very polite in asking us not to give 2026 guidance on this call, and we'll handle that in February. As I said earlier when answering something else, we've made tremendous headway on our long-term margin goals even as the platform scaled up to 135 million monthly actives, and our guidance for this year points to bookings of almost $1.2 billion. So everyone can see we're capable of growing the business, running it with discipline, and widening margins. That's a positive thing, and we intend to keep it going. We'll carry on operating with discipline and expanding the business in the direction Luis described, hopefully reaching an entirely different order of magnitude. Even so, we aren't hesitant to put money in. Our view is that profitable operations remain achievable alongside continued investment. That said, pushing for steady, straight-line margin improvement isn't something we'll put first going forward when a better option exists. The chance in front of us is simply too large, so putting money into growth takes precedence—that's the way we see it.

Ygal Arounian analyst
#49

Understood. That's genuinely useful. Appreciate the time.

Andrew Boone analyst
#50

Thank you, everyone. Thanks for fielding our questions. I'd like to dig into something we last touched on perhaps 2 or 3 years back — the idea of advanced English learners, and essentially how better efficacy pulls those learners onto the platform, plus whether AI sped that up, whether that thesis played out, and whether any deceleration in growth is affecting that group of learners. Or is there anything else you could share regarding that cohort?

Luis von Ahn executive
#51

Yeah. Like we said, I'm not sure exactly when it was, a couple of years back, we began discussing this. Teaching English represents a huge opportunity. After all, 80% of language learners globally are studying English. So we began putting money into instructing English learners better. That's what we've done. We carried out a big launch over the course of this year. Today, every one of our English courses reaches up to Duolingo Score 130, the level at which you're able to land a job — a knowledge job — using that language. So that's in place. And we've kept enhancing how effectively people pick up English. We can see it in the metrics. The count of English learners, and definitely the count of advanced English learners, has risen consistently. Something else worth noting: the fastest-growing regions right now happen to be English learning regions, so Asia stands out as a great illustration — we've definitely kept posting, or never halted posting, unhinged content there on the social media front, but beyond that we're simply gaining far more users in that area, so those are the segments expanding quickest, and I'm quite pleased with how things are going there. We did say this would take time, and it is indeed taking time for the message to truly spread that Duolingo is extremely strong at advanced English, so although I think we've achieved solid progress on that front, we haven't fully arrived yet. This is something that even across next year, we'll be watching the number of advanced English learners we have go up.

Andrew Boone analyst
#52

Could you also give us the latest on Family Plan? I understand you folks had been introducing quite a few features, plus there's the wider uptake question. What's the situation now, and how far could this go? Thanks, everyone.

Matthew Skaruppa executive
#53

Yep. Sure. Regarding wider uptake, Family Plan is still performing strongly. In the latest quarter, I believe it accounted for roughly 29% of subscribers. So it has expanded at a good pace, and there are plenty of grounds to expect that this nice growth will keep going. Take last year — in Q4 of last year, essentially everything went our way, and that's the reason our growth was so rapid then, I believe 42% year over year last year. One factor there was how strongly Family Plan performed during the New Year's promo. So there are grounds to expect we still have room to run with Family Plan, Andrew.

Andrew Boone analyst
#54

Thanks.

Justin Patterson analyst
#55

Hello, Justin. Hey there, Justin. I'm curious to learn more regarding takeaways from Energy. Around 90 days back, there was plenty of enthusiasm for this product; therefore, I'd appreciate knowing how that has developed and could be influencing some of the patterns we're discussing for this quarter; and after that, as we consider the trajectory of reacceleration, what portion of this truly hinges on perfecting the product as opposed to genuinely connecting with that cultural relevance Duolingo possessed earlier this year before hitting that obstacle involving social backlash? Many thanks.

Luis von Ahn executive
#56

All right. Regarding energy. Energy has us quite pleased; it accomplished precisely what we intended. Bookings went up along with DAUs; therefore, that was positive. Essentially, the sequence was that energy was released for iOS users initially, followed by Android. We introduced it gradually across a period of a couple of months for every platform. As of now, that process is complete. I'm unsure whether it was a couple of weeks back. Anyone who has updated the app within the last, I don't know, six-to-nine months now has energy available. And we're delighted since it performed just as we anticipated. And it has been released; now, concerning your question about product as opposed to marketing. Well, for our growth, both product and marketing matter. What's more long-term is having a product that retains users extremely well, one that also instructs effectively—that's the optimal approach. And that's where the bulk of our work goes. And we'll keep directing the bulk of our work there. Sure, cultural relevance is significant. Yet for me, it merely speeds up something where the product serves as the main dish. Regarding being culturally relevant and so on, we truly are noticing a full resurgence on that front. I don't know whether you've observed, say, everything that occurred during Halloween. I mean, across the globe, there certainly were thousands of Halloween costumes that amounted to nothing but Duolingo outfits. Thus we notice that we're witnessing quite a lot of it. And so that doesn't especially concern us.

Matthew Skaruppa executive
#57

Okay, so the one other thing I would mention there, Justin, is something we discussed during the previous call — namely, we've seen some positive results, especially in the U.S., from putting somewhat more money into actual marketing and getting a solid return there, particularly in the U.S. That took place in Q3, and while the amounts are fairly modest, we intend to keep doing it through Q4, since U.S. growth, as we've said previously, trails the rest of the world — partly due to the fact that the rest of the world is expanding very quickly — but that pattern remains, so that's simply the additional piece we would put into the marketing mix.

Justin Patterson analyst
#58

All right. Thank you both.

Luis von Ahn executive
#59

Hi, Shweta.

Shweta Khajuria analyst
#60

Hello, Luis. Hello, Matt. I appreciate you answering my question. I have two, actually. The first concerns China. Could you discuss how engagement has performed there during the quarter and across the year, plus what you anticipate and what you're observing, and also whether you might say anything about the chances for Max there, whether in this year or the next? Then the second relates to renewal rates specifically for Max. Would you describe how those moved in Q3 up to now and what you expect? Thank you.

Luis von Ahn executive
#61

Yes. All right. I'll handle the China question, and then Matt can address the renewal rates. Okay. China. Naturally, we're performing quite strongly in China. As I understand it, that's the nation where we're expanding quickest; measured by DAUs it ranks as our number two nation and is rising rapidly. It remains a fairly small share of our overall business. Right now it represents roughly five% to six% of our business, yet it is expanding, and expanding quite rapidly. Engagement there is extremely strong. Retention is strong. At present Max is undergoing testing in China. We hold authorization to test it in China. Since this is an LLM, authorization is required in order to test it. Testing is underway. Launch will occur within some number of weeks or months; I can't give you an exact figure since it hinges on approvals, though things have progressed rather well, so what I'll say at a very broad level regarding China is that it's a fairly significant opportunity for us. Naturally, though, China carries risk given that geopolitics and so on are involved, so our approach is that we aren't spending an insane amount, say on marketing in China; our spending there is quite modest, yet we simply happen to be growing considerably, so that's good, and it's a genuinely good opportunity. However, should something occur at any moment, we simply never invested all that heavily there. That's roughly our approach, and now I'll let Matt discuss Max renewals.

Matthew Skaruppa executive
#62

Okay, so setting the scene a bit — Max renewals do matter. That said, I believe someone raised a question earlier about retention across the platform more broadly. That's still holding up well. Nothing meaningful has shifted there. Then, specifically regarding Max, on the prior call we noted that in Q3 and afterward in Q4, bigger cohorts would begin moving through the Max renewal cycle. What we've observed is that Max at this point — and again, these numbers are still modest since they've been building up — was actually renewing a touch above Super. That said, it's still early. And Q4 represents a fairly sizable cohort. So we'll revisit this topic on the following call.

Shweta Khajuria analyst
#63

All right, appreciate it. Thanks, Luis. Thanks, Matt.

Luis von Ahn executive
#64

I appreciate that.

Hanyi Cai analyst
#65

Hey there, Luis. Hey, Luis. Great to have you on the line, and it's truly thrilling to learn that Max is at last being launched in China. I want to take that question a bit further, since when you look at geography, China stood out as the country with the quickest expansion during the last quarter, and regarding this quarter, you mentioned adding more users; so where do you see things going geographically? Which nation, or which area, do you believe holds the greatest room to expand during the coming quarter? Then my other question ties to the Duolingo Score that you released at this year's Duocon, since that concerned efficacy, so I'm curious: in your view, how will this sort of efficacy measurement connect to the main user performance metrics? Will it serve as the primary driver for you to add users in the coming quarter? Yeah.

Luis von Ahn executive
#66

Yeah. Thanks for asking. So when it comes to regions, Asia is the region expanding most quickly for us; and we anticipate that will hold for a while. My expectation is that will hold. The future is unknown to me, but my expectation is that will hold for a while. And without a doubt, China is at the front of that. Yet it isn't only China. Essentially the whole of Asia is expanding quite rapidly. As for the score, we're really thrilled about it. What we announced was that, to start with, every one of our main language courses now carries the Duolingo Score. And on top of that, you're able to share the Duolingo Score on LinkedIn. We're noticing a fairly strong number of people posting their score on LinkedIn. So that genuinely indicates they're using it in a sense for job reasons. And we've got that, the score right there. That score also sits within the same band as our Duolingo English Test, at least when it comes to English learners, and English learners are the group that would care most about a score of that kind. In particular, English learners across Asia are the ones who would care most about a score, so we're really thrilled about it, and its ultimate aim is to turn into the proficiency standard for at least the main languages and definitely for English, so that instead of when someone asks how much French you know or how much English you know, at the point where people say, oh, I'm intermediate, we'd like people to say, I am a Duolingo 60 in French or I'm a Duolingo 80 in English. That's our goal, and we believe we're making fairly significant headway. When it comes to the score, especially for English, the Duolingo English Test is now accepted by more than 6,000 educational institutions around the globe, among them every Ivy League university and also 99 of the top 100 universities in the United States accept the Duolingo English Test. So our belief is that pairing that prestige with the substantial scale we've built in the app, along with features such as LinkedIn sharing, should ideally position us as the benchmark for proficiency.

Nathan Feather analyst
#67

All right, thanks. And one point is that in China, our usage of LinkedIn isn't particularly high. So our aim is really to link it up with a different social media platform there.

Luis von Ahn executive
#68

That's something we're actively developing. I'm not able to share specifics about it, but it is in progress; having the score appear only on LinkedIn isn't the plan. That's what we're working on.

Nathan Feather analyst
#69

All right. All right. Thanks, Luis.

Luis von Ahn executive
#70

Thank you, operator. I simply want to express appreciation to all of you for being with us, and we can't wait to have you with us 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.