Duolingo, Inc.

Good evening, everyone, and welcome to Duolingo's third quarter 2024 earnings webcast. Today, after market close, we released this quarter's shareholder letter, a copy of which you can find on our IR website at investors.duolingo.com. On today's call, we have Luis Monon, our co-founder and CEO, and Matt Scarupa, our CFO. We'll begin with some brief remarks before taking questions. Please note that this evening's event is being recorded and all attendees are in listen-only mode. Quick reminder that we'll make some forward-looking statements regarding future events and financial performance, which are subject to material risks and uncertainties. Some of these are outlined in the risk factors of our filings with the SEC. These forward-looking statements are based on assumptions that we believe to be reasonable as of today, and we have no obligation to update these statements as a result of new information or future events. Additionally, we'll present both GAAP and non-GAAP measures on today's call. These non-GAAP measures are not intended to be considered in isolation from a substitute for or superior to our gap results, and we encourage you to consider all measures when analyzing our performance. And now, I will turn it over to Louis.

Thanks, Debbie. I guess. Hi, everyone. So, Louis asked me to cover for him. Lucky me. Anyway, let's just get this over with. How'd we do this quarter? Pretty good, I'd say. We did way better than expected in all the important metrics. Since we did so well and we feel good about next quarter, we're raising our full year guidance. Matt's going to get into the details in a minute. He's into that kind of thing. And, yeah, we're hitting our goals. Daily active users up 54% year over year, which, okay, is pretty neat, especially since users accelerated last year. Family Plan has grown to 21% of subs compared to the 18% we had at the end of last year. Oh, and our new Duolingo Max feature video call lets learners chat with me. Lucky them. Don't worry. I won't judge. Much. Maybe just an eye roll here and there. How did we pull this off? You know, the usual stuff. Product improvements and social marketing. It just works. And the grand finale, Generative AI and Automation. Yeah, that's the future, I guess. I mean, look at me, an animated character running this call. AI is going to help us be more efficient and launch products faster. Pretty cool, right? So we're investing in that stuff. Anyway, I'm done. Over to Matt. He'll probably try to sound more excited.

Thanks, Lily. I'll do my best. Now you may be wondering why we had Lily step in for Louise. Well, we wanted to give an example of how generative AI is positively impacting more and more aspects of our business. We're using it to make our product more fun, engaging and effective through features like video call with Lily. And we're using it to automate internal processes like content creation. By the way, that video only took about seven minutes to create because of the tools and the infrastructure that we put in place. Now on to the numbers. As Lily highlighted, Q3 was a strong quarter. DAU grew by 54% year-over-year, which is impressive considering we're lapping last year's 60% plus growth. Bookings and revenue grew 38% and 40% year-over-year respectively, which came even as we lapped tougher comps. And we posted an adjusted EBITDA margin of 24.7%. This quarter's outperformance was driven, in part, by the strength we're seeing in Duolingo MAXX, As a reminder, Max is our highest subscription tier, and it now includes our new AI-powered video call feature. We executed well and rolled out Max faster than we expected. Max is now available to roughly half of our DAU, and we expect this will increase by about 10 points or so as we scale it to more users, primarily on Android, by the end of the year. We also saw early signs of strong demand for video call with Lilly. We find that when we introduce new features, we see a one-time bookings gain that eventually settled down into a more predictable run rate. Some of the max bookings increased this quarter was likely driven by this type of effect. Looking ahead, we're raising full-year guidance. We're guiding to bookings and revenue growth of about 36% and 40%, respectively, for the full year. And our Q4 guide takes into account video calls' estimated impact, and our experiments around our New Year's promotion. Our Q4 guide has about 100 basis points of sequential quarter-over-quarter decline in gross margin, which is due to higher Gen AI and amortization costs related to scaling Max and its video call feature. As a reminder, our Duolingo Max tier yields more gross profit dollars, but a lower gross margin percentage than our Super tier. We're also raising our 2024 adjusted EBITDA margin guidance to 25.5% at the midpoint, which is roughly eight points higher than 2023. As we continue to make progress towards our long-term target range of 30 to 35% adjusted EBITDA. For Q4, our adjusted EBITDA guide of 24.4% at the midpoint reflects quarter over quarter operating leverage of about 70 basis points for R&D, 90 basis points for S&M, and that's offset by some slight deleverage in G&A. We ended Q3 with approximately 49.6 million fully diluted shares outstanding using the quarter-end close price, and we expect net dilution of little more than 1%, similar to last year. And good timing. It looks like Luis is back.

Thanks, Matt. I'm glad Lily was able to cover for me. Now let's take some questions, and Debbie will manage the queue.

All right. Sounds good. If you have a question, you can use the raise hand feature. So your first question comes from Justin Patterson at KeyBank.

All right. Thank you. And, Luis, since you subbed in with Lily, I will sub in with baby Luca here. He's just out for his walk and nap right now. Nothing puts him to sleep faster than earnings calls. But, you know, if I project from your current trends and think about how next year or even 2026 goes, you're probably crossing 10 million subs within about an eight to nine year period since you started monetizing. So if you think ahead, what does it really take for you to add the next 10 million, reach 20 million subs? Is that just expanding price tiers? Is that more marketing, new apps monetizing? We'll have to hear about how you're thinking about that philosophically. Thank you.

Yeah, that's a great question. I mean, you know, generally we're going to continue doing what we've been doing because it's working really well. um and in terms of monetization you know there's there's a few things that we're doing first we're growing the number of users just this is paying and non-paying users as we grow that we grow number of cells because a certain fraction of them subscribe we're going to be doing everything that we do for that that is making the product more engaging also doing our social marketing so that should increase number of users and and as you saw our daus are growing been growing between 50% and 60% really for the last two years. So there's that. And then we're just going to get better at converting these users. And there's a number of things that we can do. For one, we're going to have our bread and butter stuff that just basically makes improvements to our purchase page or when we give the offer to the users of what to subscribe. But also the new plan, MAX, We think that there's a lot of geographies where Max is going to be more interesting than Super. And in particular, for English learners, they're very interested in practicing conversation, which is what we offer with Max, the video call feature with Lily. So we think there's going to be a lot of subscribers coming from that. But generally, that's kind of the philosophy for it.

All right, Justin, thank you. And we're going to go to the next question, which is Ryan from Needham & Co.

Hi, congrats on a great quarter. Thanks for taking the question. Maybe just on the – in the shareholder letter, Luis, you talked about sort of 2 million DAUs now that are learning sort of the – using the platform for intermediate English learning or higher learning. Obviously, you talked about sort of this initiative of driving more English learners to the platform. I think it was about six months ago now. Can you just give us a sense of, on that two million, sort of what sort of progress has been made within that last six months since you've been focused on it, and maybe what regions you're seeing sort of, or languages you're seeing sort of the most attractive pickup or most near-term pickup on usage? Thanks.

Yeah, thank you for that question, Ryan. So as we have been saying, English learners are a pretty major opportunity for us. The reason for that is because if you look at the broader language learning market outside of Duolingo, the majority of the spend, about 80% of it, is from people who are learning English. But if you look at Duolingo, the amount of revenue that we make from people who are learning English is significantly less than 50%. So there's a major opportunity there. And the reason that we are underrepresented within the English learners is because we haven't historically had intermediate or advanced content in English. We started working on that a few years ago, and by now all of that content is there. So we're very happy with the content. We're obviously going to continue improving it because we're always improving everything, but the content is at least there. The other thing that we've done is we've worked a lot on placing users. English is a unique language in that because it's the lingua franca for the whole world and because in most countries people learn some amount of English in schools, Mostly when they come to the app, they have some previous knowledge, unlike most other languages where usually when they come to the app, they're beginners. So we have to really do a good job of placing users in the right place. And we've been working on that quite a bit. And at this point, we feel pretty good on the product in terms of the content, and then also the placing. So that feels pretty good, and we're going to continue seeing growth. Like we said in the letter, we have over 2 million daily active users in English in advanced or intermediate content. We think that's good. You know, that's growing pretty fast. And now one thing to call out here is that most of our growth comes from word of mouth. I mean, we're going to be doing some marketing to get the word out, et cetera, but most of our growth comes from word of mouth. Now, word of mouth growth is excellent in that it's very cheap. We don't have to pay for it. People just tell their friends. But it's also not super fast. So we expect that because historically there just hasn't – people haven't really thought that Duolingo has intermediate or advanced English. We expect that it's going to take a couple of years for really the word to get around that – our English courses are good for intermediate and advanced speakers. So that's kind of what's going to happen, but we're very excited, and that's a pretty major opportunity.

Maybe as a follow-up, it seems like as you continue to roll out new functionality that you're maybe starting or expanding to the Android user base maybe a little bit more quickly than you have in the past. Just curious how the uptake on Android users has been for the new functionality, and if you're seeing any differences in terms of how those users are converting relative to maybe in the past couple of years. Thanks.

Yeah, I mean, the way we develop, usually we put features first on iPhones and then on Android phones, where we usually call it three to six months behind on Android. And we're getting better at that. It's getting faster and faster. Right now, we are working on, for example, adding all the Macs features to Android. That's something that we're adding. One of the things to mention is that we expect, you know, usually, by the way, Android users just don't monetize as well as iPhone users. That's industry-wide. This is not just Duolingo. It usually is the case that they probably have lower purchasing power. But one thing that is interesting is we are seeing in our max features, and in particular the key max feature of video call, we're seeing that English learners are using it about twice as much as people who are learning other languages. So we expect, because Android in particular has more English learners than iPhone, we expect that there's going to be a really good uptake there.

Excellent. Thanks for taking my questions.

Thank you.

Okay. Next question comes from Aaron Kessler at Seaport Research.

Great. Thanks, guys. A couple questions. Maybe just as a follow-up on the intermediate English learning, can you just go through kind of maybe the marketing strategy to engage English new users and make them aware of that. And second, I think last quarter you mentioned pretty strong international growth, including Japan's, with benefiting from kind of more country managers. Just any updates on some of that international growth this quarter as well? Thank you.

Yeah, thank you. So, you know, we do a very specific type of marketing. You know, you've seen our social media. It's pretty unhinged. That type of marketing, our bread-and-butter marketing, is not particularly applicable to convincing people that we have intermediate English content. I mean, the owl breaking things and doing dances is not really applicable. What we do think is applicable is having influencers. We use influencers in a lot of countries, and so we're going to be doing that. It actually works pretty well. And not just us doing the marketing. For example, there's videos that people do without us even paying that do this very well. For example, there was a video recently that got a lot of views, like many, many millions of views of a guy who just – just used Duolingo to learn for a few months and then just went to Russia, and he just recorded himself trying to do stuff, and it turned out that he was able to do it. That type of stuff is really good for efficacy because it's like, oh, wow, you can go to Russia and actually get around. So that's the type of stuff we're going to be leaning in on our marketing. And in terms of, okay, so one last thing I'll say about the marketing, which is I'll reiterate what I said in the previous answer. That's what we're going to do for the marketing. But once again, the main way in which we expect that the word to get around that we have intermediate and advanced content is through word of mouth. because that has worked so well for us. So that's probably going to be the main way. So there's that. Then for your second question in terms of international growth, we're very happy with the results so far for international growth. If you ask about our DAU growth right now, this quarter we reported 54%. It's broad range. So every single country is growing well. Of course, some countries are growing a little more than others. But basically every region is growing. We have country marketing managers on a number of countries. We're about to add new ones. We're about to add Italy and Turkey. And we have ones in Korea, in Japan, in China, in Brazil, et cetera. We have a lot of them. And that works really well. And typically what it is is we hire one or two individuals there in-country, and then they help with our basically localizing our humor and our kind of TikTok and YouTube accounts. And so far, every single country where we've applied this, it has worked.

Great. Thank you.

Okay. Next question comes from Ralph Shackert at William Blair.

Good afternoon. Thanks for taking the question. Luis, maybe just get an update on the macro. You know, some investors might see this business model sort of fairly discretionary, and, you know, you're growing 50-plus percent in DAOs. So maybe just sort of riff a little bit, if you can, on why you think the model's held up so well and, you know, why you think you can continue to grow at these growth rates. And I'll ask Matt a question since he's kind of quiet over there at next.

I mean, in general, you know, we've – We see a lot of opportunities still to grow. The number of people who are learning a language in the world is about 2 billion. We have about 100 million monthly active users, so there's a lot of runway in there. And so we're just going to continue growing. I mean, by the way, like we said last time, we expect our DAU growth for the rest of the year certainly to be around 50%. So we're going to continue strong growth. And like I said, for the last two years, it's been 50% to 60% year-on-year. So we just expect that to continue happening. And over the longer term, we expect other subjects to start helping, too. So this is language learning, but we expect that math and music are going to start contributing more and more over the next few years. So we feel pretty good about Macro in that respect.

Yeah, well, Ralph, before you ask me the question, the addition I'd add to Luis is just that, you know, our entry-level price for Duolingo is free, so that helps. And then the actual subscription is just not very expensive.

uh on a monthly basis yeah even super or max so great and then just maybe matt on margins you know just kind of remind us of your framework for letting the business scale versus reinvestment opportunities obviously the models scaling pretty significantly thrown off a lot of cash and margin but how you think about incremental margins going forward um no it's a great question and i'm it gives me a chance just to kind of highlight the fact that we have scaled

incredibly well. I mean, at the top line, bookings CAGR over the last three years is, you know, 42%, including our most recent guide. While we've done that eight quarters ago, we had $2 million in Q4 of adjusted EBITDA, and we just guided to 49.7 at the midpoint. So 25x on adjusted EBITDA. So It feels like we're scaling both growth and profitability really nicely, and we're doing that with your point around incremental margins above our long-term target of 30% to 35%, and we think that that's reasonable. But first and foremost, we feel like there's a ton of opportunity. The first question was how do you get to 20 million subscribers? You had a question about macro. You got questions about DAU growth. There's lots of opportunities and lots of ways to win, so we're going to continue to invest first back into R&D because that is our primary goal. a way to grow through word of mouth. So, you know, I think it's a yes and, Ralph. We're going to continue to try to invest for growth, but we're going to continue to scale profitably as well. Awesome. Thanks, Luis. Thanks, Matt.

Great. Thanks, Ralph. Next question comes from Brian Smilick of J.P. Morgan.

Great. Thanks for taking the questions. I guess just to start on Max, you know, available to 50% of DAUs, You know, up from 15 percent last quarter. Can you just help us understand the conversion cycle of a max subscriber for when it becomes available to when they go down the funnel? And I guess, like, how does that differ versus super?

So there's a couple of things to say. Yes, we've been increasing max. The way we've been increasing max, by the way, is by adding it to more countries and in more courses. At this point, we're at around 50% of our daily active users have access to it. Now, that doesn't mean they bought it, but they have access to buy it. By the end of the year, it'll be somewhere between 60% and 70%. One important thing to say about that, by the way, is that We're adding countries, but the countries at this point, it's mainly long tail that's left. All the kind of wealthy countries we've added. Now, in terms of how the funnel is, we're seeing a lot of people who are buying Macs the first time they subscribe to anything. So they're a free user and just they go directly to Macs. And we're also seeing a bunch of people who were super subscribers who were able to upgrade to Macs. So we're seeing both of these. And It's a bit early to know exactly where this is going to settle because some of the features in Macs are very new, particularly the feature of late that has really unlocked a lot of purchases in Mac. Kind of the key feature is video call with Lily. And that feature has only been around for two months. So I don't really know exactly what's going to happen, but I get a sense that This is going to vary geographically, so there will be some countries, like the U.S., I'm guessing, where super is going to be the main package because it's a package for convenience, whereas there will be some countries, particularly the English-learning ones, main package because this is not – usually in the poor countries, people don't pay for convenience, but they do pay for things like, you know, learning how to converse better. And so there may be some cases, some countries where actually we'll have more max packages than super packages. We just don't know exactly how it's going to be yet, but that's kind of what I think will happen.

Awesome. Thank you. And I guess one more if I could just on DAU growth. You know, as we go into 2025, can you just walk us through the building blocks, of DAU growth? I mean, you know, coming off strong growth in the 50%, I mean, where do you think realistically, you know, the TAM expansion is going to come from? Will it be more English in 25 or are we still talking more non-English learning?

I think we're going to see growth from all regions. I mean, the nice thing about our growth so far is that really there is no country where we are growing very slowly. It's just they're all kind of growing pretty fast. You know, where we're going to see next year, I think we're going to continue seeing growth in our more penetrated markets like the U.S. And I think we're going to be layering a lot more English growth. So it's going to be both. And, again, the reason that we believe that we can continue growing fast is that we really are just scratching the market for language learning. I mean, there's 2 billion people learning. And in addition to that, not only is there 2 billion people learning, in some countries we are completely growing the market. I mean, for example, in the United States, about 80% of our users are people who were not in the market before. They weren't learning a language. So we just don't see that anything is – there's nothing that we see that is pointing us to some, like, cap or anything like that.

Thank you both.

Okay. Thank you, Brian.

Next question comes from Ross Sandler at Barclays.

Great. So I guess it's kind of a – Question for both of you guys. But, Luis, you were on the Decoder podcast recently. Nice job, by the way. And I think you had said that at the subscriber level that max is accretive to margin. And, Matt, you just mentioned that COGS is going to go up because of the video call feature, which makes sense. But could you guys just make sure we got those right, just unit economics versus overall cost? And then I guess more importantly, Luis – Inference costs are down like 90% in AI right now. And so as that gets cheaper and cheaper in subsequent years, how does that potentially change either the max rollout or feature rollout, more things like the video call with Lily, et cetera? How are you thinking about that broadly?

Yeah, let me answer your second question first, and then I'll let Matt answer the first question. So, yes, costs will continue going down. That's our expectation. And not only is cost going to continue going down because large language models are going to be cheaper to query, it's also the case that we ourselves have not spent a lot of time optimizing costs. I mean, the... the directive that all our teams have is don't worry too much about cost at the moment for large language models because that's going to naturally go down. So just develop the best features you can. And over time, we're going to, you know, if we see that maybe the LLMs are not going down in cost all that much or something, we're going to start optimizing ourselves. So we really do believe that cost will go down. What that will do is it will allow us to offer things like video call with Lily in – at more reasonable prices for certain countries. We think that the price is good for a country like the United States. We don't think that the price at the moment – I mean, we don't even have it in India, but at some point we're going to have it in India. But we probably at the moment cannot offer it for that attractive of a price. But I think sometime next year the price will be pretty attractive. So it'll just allow us to offer it to people in poor countries. Now, the thing about people in poor countries is those are exactly the ones that want to learn English, and the people who want to learn English are exactly the ones who want to more so practice their conversation. So we think this is going to be a pretty major unlock, the decrease in price. Yeah.

Yeah, and just to follow up on the first part, Ross, you know, we launched Max because we – had a belief that there was demand for Duolingo at a higher price. And Gen AI enabled us to add features to that tier that allowed us to charge that price. And, you know, for example, like Luis said, in the U.S., it's, you know, roughly 2x the price of Super. So when you have that 2x the price, you have plenty of gross profit dollars to play with. And so that's when Luis was on the podcast, we definitely make more gross profit dollars per subscriber for Macs than super. The margin, the gross margin percentage is lower because in addition to the app store fees, there's a set of LLM calls. But, you know, for a free cash loan, it is a deep dollars. We're making more. And so we still feel good about it. The point that Luis was making on cost also applies to price. You know, we have charged the team with just making a wonderful product. And you saw an example of that on the or something like that on the beginning of this call, we still haven't optimized all the way pricing and costs. And so we'll do that over time, and, you know, that will help margins as well.

Okay. Next question comes from Andrew Boone at JMP.

Thanks so much, guys, for taking the question. I wanted to go back to kind of product testing and curve. Right. And so if I think about tests as a main driver of what is retention on the platform, you guys have basically doubled the user base for the last two years. But that's also with two years more product testing that's now behind you. Can you talk about what is the pace of testing and whether that's faster today and whether you guys are seeing more gains, getting that larger base of users and faster testing? Or how do I think about those offsetting features?

Yeah, the good news is we're running – the number of tests that we're running per quarter is increasing. So we are running more tests per quarter. The success rate for the test is about the same, and it's actually a funny thing. It's about almost exactly 50%. So every test that we run has about a 50% chance of succeeding. And by succeeding, I mean that it did what we wanted it to do, like it increased whatever metric we wanted it to do. to increase. So the pace of tests is increasing, and we feel pretty good about the number of things that we have coming up over the next, you know, we have visibility. I have visibility for about six months. We feel pretty good about the things that we have for the next six months in terms of number of tests.

Yeah, and Andrew, the other thing I would add is just that the teams do run experiments from time to time to try to enable themselves to speed up their own experimental frequency, to get data faster, to analyze the experiments more rapidly. And so we're always trying to run those, and we're going to run some this quarter, for example. So we think things can speed up.

That's helpful. And then if I think about a similar question within the framework of what is recently lost products, if I think about adventures and video calls, can you guys maybe benchmark that as a surface area for testing versus what is streaks or maybe the leaderboard in terms of past products that have been successful in terms of driving, again, Kerr? Thanks so much, guys.

I guess maybe it's taking a step back a little bit. The way we operate for the Duolingo app is we have a number of different areas that cover basically the main things we want to improve. And there's three main things we want to improve. We want to monetize better, we want to make it more engaging, and we want to teach better. And depending on what the feature is related to that area is going to be working on. So, for example, the streak is something that makes the app more engaging. And we are running, and we have run I don't know how many experiments on the streak, hundreds, possibly thousands, and we have way more that are coming up in the next several months. Teachers like Adventure and Video Call, these are usually, the main goal of those is to teach better. So they're being run from the areas that have to do with teaching better. And they're going to continue running a bunch of experiments. And the last thing that I'll say is the number of people that we have working on each one of these kind of engagement, teaching better, and monetization is roughly the same. So we have about equal number of people working on each one of them.

Okay. Next question comes from Curtis Nagel at B of A. Great.

Thanks very much for taking the question. I guess the first one, maybe just focusing on the family plan, it picked up a little bit at, I think you said, 21% of subs. I guess in terms of the contribution of some of the new product features you've added versus or maybe in addition to, you know, trying to increase visibility of the product, you know, what's been the contribution in those two? And where do you think this – you know, this year can evolve in terms of total mix, let's just say, over the next 12 months.

So we've done two things for the family plan, and you're right. One is we've made the family plan better. I think we've added features to it or fixed some things that weren't exactly bugs for things like, you know, kids under 13, the parents couldn't see their name. Now they can. So we've done stuff like that, and then we've also made it more visible. The majority of the contribution is actually making it more visible. You know, the features are good, and that will probably help over the long term in terms of the retention of the family plan, but just generally making it more visible is what has had the most contribution. In terms of what exactly the penetration will be over the next year, it's very hard to say. I mean, I honestly just don't know the answer to that. I expect it to continue going up, but I just don't know at what speed.

Got it. Okay, and then just to follow up, I don't think we touched, maybe it was in the third letter, I missed it, but the resurrected user, which, you know, was the focus for the last quarter, making some progress, and you're focusing there. So, in terms of however you want to measure it, you know, whether it's subs or DAUs, the contribution in 3Q relative to 2Q, how did that look?

Yeah. Resurrected users are a pretty big opportunity for us. What's happening as time goes on – The way we look at top of funnel is basically either users that are coming in brand new to the app or users that are coming back after a long hiatus. That's top of funnel. So one is new users, the other one is resurrected users. Obviously, when you just launch an app, all of your users are new users because you're brand new. As time goes on, a higher and higher fraction of your users that are from the top of the funnel are resurrected users. At this point, we have gotten to the point where more than half of the top of funnel is resurrected users. So on a given day, a larger number of users are coming back to the app after a long hiatus than new users. And that's just natural for a product. Of course, this also varies per geography depending on how penetrated this is, the case. I mean, in the United States where we've been operating for longer, there's a higher fraction of resurrected users versus in a country like maybe India where we just haven't been operating for that long. This is a pretty major opportunity for us because we just haven't worked too much on the experience of when people come back. So we feel pretty good about that, and that's going to be a good area of growth. The other thing that I should mention about resurrected users is a thing that is important for people to understand. A lot of times when people think about these subscription models, they think, well, after a user has left, they'll never come back. That is just not the case with Duolingo. I mean, with Duolingo, it is rare to see that a user leaves and never comes back. Usually, when they stop using Duolingo, they usually come back, you know, I don't know exactly how long it will take them, but three months later, two years later, et cetera, that's a pretty common thing. And interestingly, this is just an interesting tidbit, the most common answer when you ask users why they stop using Duolingo is, I haven't stopped. which is a funny thing. They still see themselves as Duolingo users, even though they may not have been around for 35 days or 40 days.

Appreciate the comments. Thanks.

And this question comes from Wyatt Swanson at D.A. Davidson.

Hey, guys. Thank you for the question. Could you discuss maybe what you're seeing in terms of overall subscriber retention, just like any changes in retention trends as you roll out additional features? And then are you seeing any differences in subscriber retention for Superplan versus Max? Yeah.

No, I appreciate the question. You know, the overall retention hasn't changed since we last talked about it, you know, last quarter or the quarter before. blended retention on the platform is pretty stable at this point. That could change over time. We don't actually have enough data yet on Macs at scale to really know how that's going to retain. We have enough data that makes us feel comfortable that the LTV will be superior to super. But we don't know exactly where that's going to stabilize, just like we don't know where the top line run rate will stabilize. So you know, we're watching and we'll let you all know how it's trending, but so far there's been no major changes on the platform and retention. Got it. Okay. Thank you.

Okay, next question comes from Arvind Ramnani at Piper Sandler.

Hey, thanks for taking my question. By the way, really cool, cool animation. Yeah, Luis, I really like listening to you. This was really very enjoyable.

You know what? She's doing part of my job now. Over time, she's going to do more and more of my job, and I can just retire. Yeah.

I don't know the core part of her job, but at least, you know, sort of, like, it was just very enjoyable to see that. Probably one of the most enjoyable earnings calls I've had in several quarters. Good. Yeah. Just a quick question on GAOs, right? I mean, I think it's a big focus for investors, and I think, like, we've talked about in the past, you know, like, long-term... Should we be comfortable as long as DA use is, you know, at some point higher than revenue growth? That's like a more normalized way to think about it. Because the last two years, you know, we've talked about it extensively. There was a lot of unusual factors that got it to like a really, really kind of impressive level. But like longer term, like how should we think about the relationship between revenue growth and DA use?

You know, I don't know exactly how to answer that question. I mean, my sense is for the last – it really has been a little over two years where every single quarter our DAU growth is somewhere between 50% to 60%. Now, this obviously won't last forever. I mean, we're saying for the rest of the year at least it'll be 50%. It won't last forever. But my sense is that we are going to continue seeing strong DAU growth for a while just because – the main way in which we grow is word of mouth and also adding more features or improving our features. And we just have a really good set of improvements planned. So I think that will be the case for a while. I don't know what steady state will be, and I don't even know what that steady state means, but I don't know if, Matt, you have anything about that. I'm not sure how to answer that question.

Yeah, no, I don't have the perfect ratio, Arvind, but I do think it just gives me a chance to remind everyone that we have not just several ways that we can grow users. So Luis has already talked about new to the platform, you know, resurrected users or users have been away from the platform for a month coming back. You know, there's many levers we have to grow users. And then there's many experiments, hundreds of experiments every quarter where we grow conversion from free to paid. And then there's experiments and vectors that we have to retain subscribers better. And then there's mix shift between plans. And then there's pricing. And, you know, I won't bore you, but I would just say that there's a lot of vectors throughout the funnel from free user to retained paid subscriber that give me confidence that we can sustain really nice revenue growth, you know, above 25%, say. in a lot of different DAU environments. I agree with Luis. We have a bunch of belief that we'll be able to grow users nicely for some time, but I think there's a ton of levers. So, yeah, that's why we feel comfortable that our revenue growth rate, you know, should stay strong for a long time.

Yeah. Sir, I just have one follow-up question. You know, certainly your DEOCON consumer day was really enjoyable, but, you know, We really don't get, I mean, investment company, we really don't have any insight of like, you know, how does one compare to the other, right? I mean, we're looking at from a product and things. Are there any kind of metrics or anecdotes you can share from this year's like kind of consumer day versus, Duocon this year versus last year? Any incremental like from a metrics or anecdotes?

For Duocon in particular, yeah, the number of live views on Duocon was three times the number of live views from last year. There's all kinds of ways to measure them because, by the way, live views is one thing, but we also get to see a lot of views over the next week, over the next month, et cetera. But it's generally... Really, every single year, we're just getting a significantly higher number of both live views and also historical views for each Duocon. And I think it just has to do with the fact that our brand is more well-known.

yeah so if we can just slip one last one in you know when you think of dual link of max and and super um i think one of the things has been like the what product or features go into one versus the other uh they'll have is it still like a very ambiguous line or do you have a bright line of what goes into each and then of course you have max right like which is another thing yeah yeah i mean

I can tell you where we're at at the moment. I cannot guarantee that we'll be there three months from now because things change here fast. At the moment, where we're at is super Duolingo has the features that basically where you pay for convenience. For example, turning off ads or unlimited lives, like unlimited hearts. That's paying for convenience. Max. The more time passes, the more in my head and in our product team's head is basically video call with Lily. We have other features in there, but by a wide margin at this moment, video call with Lily is the killer feature for Max. And if you think about that, what that is is just practice conversation. So at the moment, the way I'm thinking of Super versus Max is Super you pay for convenience, Max you pay for getting better at conversation. and that's it. That's at the moment. Again, maybe when we talk in three months, I'll tell you something completely different.

Perfect.

Thank you so much.

All right. Next question comes from Alex Farr of Raymond James.

Thank you. Luis, I kind of wanted to follow up on that video call conversation. What has the early data told you so far? You had a trial period, now the limited launch in terms of usage, and I'm curious if there's been any notable uptick in terms of, like, growing time per session, or better streak retention that that might translate to higher max retention or greater adoption longer term? Thanks.

Yeah, the usage metrics for video call are very good. And what I really like is that they match our expectations. And by that I mean English learners use it more than non-English learners because they're more interested in conversation. Also, for any language, more advanced users use it more than less advanced users. And it's exactly what we want. And we're seeing, you know, we're making changes to the experience. We're trying to make it more engaging. For example, you saw the video that we played at the beginning of this call. Lily has a lot of different kind of facial expressions, et cetera. We're adding more and more of those to make it more and more realistic. And every time we do that, it becomes more engaging, and we are seeing that. So we do see increases in the number of calls that people have. or the amount of time that they're spending with it whenever we make it more realistic or the conversation topics are better. The other thing is we're going to be getting a lot better with the actual conversation topics, for example, right now. lily lives in a world that where for example news doesn't really happen today what should have happened today is lily should have talked about if you call her she should have talked about the election um but that that's not something that is currently happening but in a few months that's that's how it's going to be um so we're we're feeling pretty good about that um i i hopefully i think that answers your question yeah no great color there um

Matt, maybe this is a follow-up for you, but you referenced it in the prepared remarks. It's still a little over a month away. Any changes to how you're planning on approaching holiday, kind of New Year's season from a promotional standpoint this year, given you'll have a wider max rollout, and any different kind of timing assumptions that are embedded in your outlook? Thanks.

Yep. No, the outlook is always a bit uncertain around the fact that we – you know, launch our one promo of the year, and it has four days in December, and then obviously the month of January. So that's the same as it's been in years past. We do run experiments every year to service things differently, to promote it in different ways, and we'll start running those. We usually run them in early December and then launch them. Those are a bit tougher to predict, but that's all incorporated into the guide. Great.

Thank you both.

And next question comes from Shweta Kajaria at Wolf Research.

Thanks, Debbie. Thanks for taking my questions. I guess I'll try two, please. One is on international markets. If you talked about France and Korea in last quarter, you had mentioned those two as adding marketing managers post your success in Japan. So how is that tracking and have you added more international countries with marketing managers? That's question one. And then the second question is on your marketing strategy. If you're seeing anything, if you could comment on the pricing that you're seeing on social media platforms, how have ROIs trended for you? Anything in particular that you're seeing over the past quarter and in this quarter today? Thank you.

Okay, great. So international markets, yes. We very recently added marketing managers for France and Korea. France is a little longer, maybe a few months, and Korea is really like weeks. So it's too early to say in Korea. France is looking very good. It's just, again, every time that we follow the playbook, it ends up working. It may take a little longer in some countries than others, but it ends up working after a few months. And we are already seeing it working in France. We are about to add also, like I mentioned, Italy and Turkey. So we're going to be doing that. And we have a few others that we're considering adding for next year. In terms of marketing strategy, you said things like social media prices, et cetera. We don't pay for the vast majority of our social media stuff. I mean, we do a little bit of performance marketing, but the social media that we're known for, like our TikTok videos or our YouTube shorts, et cetera, that is all organic. We don't pay for that. So the cost of each one of those videos is the cost to make it, and the majority of them, it's like a couple hundred dollars to make it, and that's it. We do a little bit of performance marketing, but it's kind affecting us and whether it doesn't really affect too much whether the prices are going up or down. You know, I'm sure there's an effect, but it's just not something that I personally track.

Thank you.

Okay, next question comes from Mark Mahaney at Evercore.

Okay, thank you. Could I run a couple questions by you? The growth of the U.S. market, you know, it's kind of as your oldest, most mature market in terms of bookings or revenue or MAUs or DAUs. Is that still pretty consistent with that of the global growth rate? That's question one. Secondly, I'm sorry if you covered it early on. Have you said anything about what kind of traction you're actually seeing for Macs? I know you talked about what percentage of the user base has access to it now, but did you provide any disclosure at all on what kind of traction you're seeing for it? And then third, I just want to ask about an ARPU question. And Matt, I know there's a lot of moving pieces in that kind of average bookings per average sub, but one of those factors is this greater adoption of super and max. Are those big enough now to kind of sustainably cause ARPU to grow going forwards? Thanks a lot.

Let me take the first two, and then Matt will take. Matt answers anything with acronyms. Okay, so the U.S. market, yes, growth in the U.S. market is good. It is growing similar to most countries. Again, it's not the case that, you know, the U.S. is the fastest-growing country, but it's also not the slowest country, and it's also not the case that we have huge disparity between countries. Most countries are kind of growing in similar rates, and it's a similar rate to our overall average. So the U.S. is somewhere, I don't know exactly where it is, but roughly average. So we feel pretty good about the U.S. growth. In terms of max traction, yes, we feel very good about max traction. And in particular, the reason we feel very good about it is because over the last couple of months, we added this extra feature, which is video call with Lilly. And that is really allowing us to advertise max in a much better way. Because what happens with video call with Lilly, the previous max features that we had, explain my answer and role play, were a little hard to understand for users. But this one is so easy to understand. Within two seconds, people get the idea. It's like, you can get to talk to Lily. And you just say that, and people are like, oh, whoa, yes, I want that. And so we're seeing really good traction in terms of people actually buying Macs. I don't think we released the precise numbers just yet, but we feel good about it. Yeah. And then, Matt, do you want to talk about ARPU?

Yeah, Mark, so I'm talking about ARPU, so actually revenue, not bookings on this part of the conversation. As you can see in the numbers, you know, our ARPU trended towards 0% year over year, which we've talked about for the past couple quarters about getting it to flat-ish, I think was a word I used. And it's now, I guess, technically flat-ish. It's 0% year over year. And we do think it can go a bit higher. You know, the things that impact ARPU for us are basically what you said. They're plan mix. So is it super? Our family historically has changed the pricing with with family plan obviously being higher ARPU. And then going forward, MAX will definitely impact that, given it's 2X the price. Right now, MAX is not at a scale that it's really showing up materially in the ARPU mix right now. Again, because revenue is amortized over 12 months, and so we've only really seen a material impact in the past quarter. So I expect that to come through. And then the other things that change ARPU are, you know, foreign currencies, regional, country mix. So I think going forward from that, you know, we'll update you all on how family plan super and max are trending on ARPU. But right now we feel good about the trend. It's gone from, you know, negative year-over-year growth to 0% this quarter and, you know, probably can stay there or go a little bit higher.

Okay. Thank you, Matt. Thank you, Luis.

Thanks, Mark. Okay, we have two more questions to get through in eight minutes. We've got Chris Kuntarich from UBS.

Great. Thanks for taking the question. Maybe just another one here on Max. I think you talked about seeing more engagement with conversations with Lily from English speakers versus not. Like, I guess it's English versus not, which puts us at kind of a 50-50 dynamic where kind of be thinking about who's actually adopting the Max sub at this point? Should we be thinking about it as roughly evenly split at this point, or are we talking about English being more closer to 2 to 1, 3 to 1 sort of adoption of Max at this point, or any sort of color you can provide to help frame that would be helpful?

Yeah, I'll give you some tips on this. Max is significantly more expensive than Super. In the U.S., for example, it's twice the price. So you would expect that wealthier countries are adopting Max more, and there's some push there. On the other hand, the main feature for Max is video call with Lily, which we are seeing that English learners, which predominantly are in poorer countries, use that twice as much when they have access to it. So that would give a push towards English learners slash poorer countries adopting Max. So there's this push. push and pull on either side. I don't know where it's going to settle because, for example, right now, Max is not available in some very large English-learning countries. It's not available in India, for example. So we're going to put it there and that will happen over time. So I don't know where it's going to settle. At the moment, there are more Max users in wealthy countries, so they're usually the ones that are not learning English. But that, I think, is going to change. I don't know if it's going to be I just don't know where it will settle, but that's going to change some. Got it.

And maybe just I noticed the call out there was a meaningful contribution from Max to total bookings in 3Q. I don't believe that was in the 2Q press release here. Any sort of kind of framework to be thinking about there? Like should we be thinking about that now as over a low single digit sort of growth contribution within the quarter?

Yeah, I mean, I think that that's why we called it out was Q3 was really the, you know, a change in terms of Max's materiality. And then as I called out in the Q4 guide, the Q4 guide does incorporate what we believe will happen with Max in video call in Q4. So it is now a meaningful part of the guide and our performance. Again, I think that we're still in the initial stages of this, especially with video call. So in Q3, I referenced the fact that that led to, like, some one-time bumps as we rolled it out more broadly. I think we'll probably see some of that in Q4 as well. But, yeah, I think that's all incorporated in the guide. Okay.

Got it. And just a clarification from earlier, did you say whether or not we should be expecting it with the New Year's discount offering?

Yeah, we're going to run experiments on it. But you'll just have to show up in the last four days of the year and figure it out.

We'll be there.

Thank you, guys.

Okay. And our final question comes from Crystal Lee at CMS.

Okay. Thank you, Benjamin, for taking my questions. I'm very honored to be the final one asking questions. So I'm just wondering how's your view on the pain ratio trend for Duolingo Max in the longer term? Do you think it will exceed that of the super Duolingo plan? And could you share more view on the competition landscape on the AI video core products? Thank you.

Yeah, we don't really know where it'll settle, whether it'll be a higher fraction of people doing max versus super. I think it'll depend on the country and our relative prices to them, so it's very hard to say where this will end. Now, in terms of competition, generally, what's gets us to stand out, what gets us to be the category leader in language learning? It's a number of things. I mean, first of all, we have understood that the hardest thing about learning something by yourself or learning a language by yourself is staying motivated. So everything we do tries to keep you engaged. I mean, the app is very gamified. You know, for example, also our conversation feature, you don't talk to a random character. You talk to Lily. and Lily starts becoming your friend, et cetera. So we really try to make everything we do engaging, and that's something that really makes us stand out. The other big thing that makes us stand out is our premium model, where really the vast majority of our users use Duolingo for free because our free tier is very good. Now, that does a lot of things. For one, it gives us a very large scale. And having such a large scale allows us to collect data to teach better and also to make it more engaging. So we just have a lot more data than anybody in history about how people learn languages. And the other thing that the free tier does is that it acts as our marketing engine. The reason we can be so efficient with marketing, I mean, we really spend very little on marketing compared to most apps of any kind. And the reason that we can do that is because our free users basically act as our marketing engine. They tell their friends. And so that's what we think differentiates us from competition. Thank you.

Okay, thank you.

That's very helpful. Okay, well, that's it for questions. I'll turn it back to Luis to wrap it up.

Thank you, Debbie. I'd just like to thank everyone for joining us. And until next time, enjoy video calling with Lily in whatever language you're learning.