Duolingo, Inc.

second quarter earnings webcast. We hope you enjoy that celebrity compilation, which complements the themes we discuss in this quarter shareholder letter, which was released today after market close. You can find that letter on our IR website at investors.duolingo.com. On today's call, we have Louise Von Ahn, our co-founder and CEO. and Matt Scaruba, our CFO. They will begin with some brief remarks before opening the call to questions. Analysts will be able to ask a question by using the raise hand feature. And please note that this event is being recorded and all attendees are in listen-only mode. Just a reminder that we'll make forward-looking statements regarding future events and financial performance, which are subject to material risks and uncertainties. Some of these risks have been set forth 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 financial 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 GAAP results, and we encourage you to consider all measures when analyzing our performance. And now I'll turn it over to Luis.

Thank you, Debbie, and welcome, everyone. I'm pleased to share that we had another record quarter. In Q2, we achieved our highest ever daily and monthly active users, revenue, profitability, and free cash flow, and we surpassed 5 million subscribers. We also just celebrated our two-year anniversary of becoming a public company, and I'm very proud that we have outperformed the lofty goals we set for ourselves at the time of the IPO. Thanks to our continued strong performance, we're raising our top line and profitability guidance for the year. Matt will walk you through our updated outlook shortly. Our strong results are a testament to our relentless focus on making our product more fun, engaging, and effective. We delight our learners who tell their friends and family about us, which drives our organic word of mouth growth. Add to that our unique and efficient, though at times unhinged, approach to marketing, and you get a brand that has become synonymous with language learning. And that creates opportunities for us to be part of cultural moments, like you saw this past month when we were referenced in the Barbie movie. I should mention that this was an inbound request to us. We didn't seek out being in the film, which I think is a reflection on the strength of our brand. Over the past eight quarters, we've seen very strong DAU growth, and that growth has been high quality and has been broad-based, with users coming from all regions of the world. The US continues to grow nicely, and some of our fastest growth has come from the wealthier European countries. This growth not only validates the huge addressable market of language learners, but because of the power of our freemium business model, which I've discussed in previous shareholder letters, strong user growth drives strong financial performance. We attract free users primarily through word of mouth, we delight them through product improvements driven by experimenting and optimizing the app, and then we convert them to paid subscribers. This playbook for growing subscribers has worked exceptionally well. Last quarter, the focus of our shareholder letter was on how AI has been part of our strategy for a long time. We have been using artificial intelligence for years to make our product more personalized and engaging. We're also embedding the recent advances in this type of technology throughout our products and the company. For example, we're using generative AI to speed up our script writing for Duolingo stories and to more efficiently scale our course content. We're using generative AI to continue to innovate on Duolingo Max, which is a higher tier offering. We will continue improving Max features and testing pricing and packaging before rolling it out more broadly, as we do with all our major changes. Using new technologies to make excellent products takes time to get right. But it's exciting to think about how the acceleration in AI can help us achieve our vision of teaching you as well as a human tutor. And with that, I'll turn it over to Matt.

Thanks, Luis. In the second quarter, we outperformed our expectations for user growth with DAU and MAU increasing 62% and 50% year over year, respectively. This took DAU to an all-time high of 21.4 million and MAU to an all-time high of 74.1 million, respectively. On our total paid subscribers, we increased them by 59% to 5.2 million. Our strong user and subscriber growth fueled our top line performance with bookings and revenue increasing 41 and 44% year over year respectively, or 42 and 46% on a constant currency basis. We continue to manage the business with cost discipline and this quarter we delivered our highest quarterly profit. Our net income totaled $3.7 million compared to a net loss of $15 million in the year ago quarter. Note that our net income benefited from a 1.3 million non-cash tax benefit. We also posted a record high adjusted EBITDA of 20.9 million, or a 16.5% adjusted EBITDA margin. Note that we moved some marketing spend from Q2 to the back half of the year, and that this increased our adjusted EBITDA this quarter by about half a point, or a point and a half. Based on our strong results and trends, we are raising our four-year guidance and issuing the following for Q3 2023. $136.5 to $139.5 million in total bookings, $129.5 to $132.5 million in revenue, and an adjusted EBITDA margin of 13% to 14%. For the full year 2023, we are raising our guidance to $569 to $575 million in total bookings, $510 to $516 million in revenue, and we are updating our adjusted EBITDA margin range to 14 to 15%. Our full year guidance calls for 33% year-over-year bookings growth and 39% year-over-year revenue growth at the midpoint. We feel confident raising our top line guidance because of our strong user growth and continued strong free to pay conversion. And because we've achieved significant operating leverage year over year across the business, that gives us confidence to raise our full year adjusted EBITDA margin guidance by about 300 basis points compared to what we issued on our last call. Let me go through how our Q3 operating expenses are expected to compare to Q2. We expect non-GAAP R&D as a percentage of revenue to increase by about a point. And we expect non-GAAP sales and marketing to increase by about two points due to the sales and marketing spend that I mentioned that we shifted from Q2 to Q3. Non-GAAP Q&A will be relatively stable as a percentage of revenue in Q3. In Q2, our average subscription revenue per subscriber declined by about 7% year-over-year, driven by foreign exchange impacts and regional pricing. At current exchange rates, we expect that the year-over-year decline in this metric has bottomed out and that the year-over-year change will approach zero by Q4 as we finish lapping FX and regional pricing impacts. Our guidance assumes current prevailing foreign exchange rates, and as a reminder, roughly half of our revenue comes from outside the U.S., so every 1% increase or decrease in the value of the dollar versus our basket of currencies has about a $1 million headwind or tailwind, respectively, on total bookings for the second half of the year. As Luis mentioned, we're excited by Generative AI's potential to help us accelerate our mission, and we are experimenting with that in Duolingo Max, our higher tier subscription. We have not yet included any material amount of bookings or revenue from Max in our guidance, and we'll keep you updated on our progress in the coming quarters. Finally, we ended the quarter with approximately 48.8 million fully diluted shares outstanding, using the quarter end closing price. We continue to expect to end the year with about 2% dilution from equity issue to employees. And with that, I'll turn it back to Luis.

Thank you, Matt. I'd like to thank our team of amazing and talented duos for all their hard work and helping us deliver another record quarter. These results would not be possible without their dedication and passion. And now we would be happy to take your questions. I'll turn it back to Debbie to manage the queue.

All right. Thanks, Luis. As I mentioned earlier, if you have a question, you can use the raise hand feature. So the first question comes from Ryan McDonald of Needham.

I've got to unmute myself. Thanks for taking my questions and congrats on an excellent quarter. Luis, I wanted to start with a commentary around social marketing and that you talked about in the shareholder letter. you've always done a really great job of being able to drive usage and users to the platform from some of these social media channels. And given that we have a new channel that's emerged really quickly in threads, I'd be curious what you're doing there from a marketing perspective and whether or not you think this can be sort of an additional tailwind to users to your platform based on those initiatives.

Well, thank you. Great question. So yes, Threads, I mean, on the day it launched, our social media team was all over it. We actually got quite a good number of engagements with it. So we're looking into it. I mean, in the end, if Threads becomes a very powerful social network, you bet we'll be there. I think right now it's significantly smaller than things like TikTok or YouTube. So we're not seeing the impact there, but we're definitely, you know, You will see us experimenting there for sure. I mean, we have an excellent social media team. You know, what I'll say in addition to this is it's important for people to understand all of this marketing is not paid for us. I mean, usually we just do all of this organically and we're going to be unhinged there too.

Excellent. Can't wait to see more from there. Second question is on the use of AI and how it's being built into the product. I'd be curious to hear some of the additional feedback of what you're hearing from customers so far that are using the Duolingo Max features. And then how did you see the AI integration evolving over time? We're starting to see newer models where maybe instead of leveraging open AI and that functionality and sort of having to bear the costs on a per interaction, that Some vendors are starting to build customized large language models for their own content, which can potentially save on costs and create some proprietary nature of the AI. But just curious how that continues to evolve for you.

Yeah, these are excellent questions. So first of all, for Max, just to remind everyone where we're at. So Max is a higher tier subscription offering. We're testing the price, but right now roughly 2x the price of Super, although you really are going to see us test the price all over the place because we're testing everything. We're testing Max on a very small fraction of our users, and this is very standard of our feature development, especially for very large features. So what we do is we start with a tiny fraction of our users, and we start making the features better and better. And as they get better and better, we start giving it to larger and larger fractions of our users. For a large thing like Macs, just as an example, the last very large change we did was whenever we changed to a new home screen. We did the path home screen. That took us about a year to roll it out from the first test to finally giving it out to our users. I don't know how long Max will take, but it is something like that. And when we first started testing Max, it had two features. One was called Role Play, which allowed users to practice conversation. And the other one was called Explain My Answer, which gives you an explanation for when you make a mistake. We've been working on making those two features better. For example, role play now is a lot more interesting. So whereas before it was probably a dry conversation where you had to maybe just order a croissant, now you may order a croissant, but something weird happens in the middle where like, I don't know, like a burglar enters, and something weird happens in the middle. So mayhem happens, and it's a much more interesting conversation. So we're improving those features. And then the other thing that we just did for Max is we added another feature, which is an in-lesson coach. Basically, before you submit your answer in a lesson, you can tap the thing, and it'll give you some hints about how to answer or give you some grammar hints about how to answer. The reception is very good. I mean, I'm very happy with it. And so far, we're very happy with it. But we're just testing our way so that by the time we give it to all our users, we will have found what we believe to be close to an optimal price and close to optimal kind of features to give it up with. But so far, so good. In terms of costs and in terms of developing our own model, for now we're using OpenAI and we're very happy with it. Our belief internally is that the cost of these models is going to go down. And so for now, we're spending most of our effort on just making the best possible features out there. rather than on saving this on this because our belief is that we we could save at any point and you know basically starting to optimize the cost of this is is what we would do after we're super happy with all the features um and we just haven't gotten there yet thanks for the keller and congrats on again on a strong quarter thank you thank you ryan thanks ryan uh next question comes from mario low at barclays

Great, thanks for taking the questions. The first one is on the average description revenue per sub. So Matt, I think you mentioned by the end of this year, it's going to be approaching like flattish growth. I'm just curious to hear your thoughts on duolingo's pricing power overall and the thought of increasing pricing on the core products in certain GS going forward, as we have seen from other consumer mobile apps.

Yeah. No, it's a great question. And Mario, you're exactly right. We expect that by the year end, your point, the year of your growth in ARPU will be essentially flattish. And then when you talk about pricing power, I think it goes back to what we've said a bunch of times on these calls, which is we experiment with price and packaging a lot around the world. So like in the US or Western Europe, raising price will be just a normal experiment. That's just a normal course thing we'll test from time to time. There are times when we do bigger pricing tests, like Q2 of last year when we did the regional pricing tests in about 100 countries at once. But we're always, I guess, fine-tuning our price. And that includes raising prices over time in certain places. So yeah, we're definitely going to consider that. And it's something that could happen in the future.

Great. Thank you. And then maybe just a high-level one. I saw on your blog that you guys had an article in terms of streaks by user age. And I think it said that users over 60, like 30% of them had a streak of over a year, whereas like users 13 and 17 was less than 5%. I guess, why is that? And any learnings that could be applied to kind of the younger users from the older generation? Thanks.

I probably don't have teenage children. It's harder to get them to do something. I mean, it just is the case that we have noticed that people who are older just are more constant in their habits. And that's that. Now, I'll say, I mean, we are very happy with our results for younger audience. I mean, we are much more well-known among younger audiences than we are with older people. It just so happens that older people are a lot more committed. I don't know if we can change that. But we're always working, just in general, not necessarily per age, we're just always working to make the product more engaging. And you can see that in a number of ways. For example, our DAU to MAU ratio keeps getting better and better. It's now about you know, 29% essentially. And it just keeps getting better and better every quarter. And the other thing that keeps getting better and better is the number of people with long streaks. So the number of people with streaks longer than a year, that just keeps, that number keeps growing and it's, you know, multiple millions. So, so yeah, I mean, just generally, we just keep making the product more engaging. Great.

Thank you.

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

Great. Thanks for taking the question. Since your IPO, you know, you're doing a very strong, consistent performance, particularly on the top line, but even more so on the bottom line, you know, this year, the new guidance is somewhere in the range of a thousand basis points increase year over year, which is exceptional. Just philosophically, how are you thinking about margins going forward, you know, without quantifying it, you know, are you willing to let them run? Will you reinvest for new products? That's the first question I have a follow-up.

Yeah. No, thanks, Ralph. And definitely agree. Since we've gone public, the top to bottom, our performance has been strong. Strong user growth, strong top line. And this year in particular, strong adjusted EBITDA performance. As we've said before, we're glad that the business is both getting more profitable, more cash generative, et cetera. And the way we think about it is we always want to invest back in the business first because the growth opportunities ahead of us are so robust. I mean, we're at the start of monetizing our own user base and monetizing a very large market. So we're going to continue to invest. That said, I think this year we're proving that we can grow nicely and become materially more profitable. So it's a yes and for us. If you're thinking about you know, your, your comment around letting margins run. If you look at other companies and, you know, our industry kind of our IPO cohort, roughly our time when we went public, you know, when they first become materially profitable, their first year of real profitability, it's usually a big step change kind of like ours is looking to be. And then after that, they make steady progress towards their long-term EBITDA targets. And I think that's more in line with what we're thinking over time. Great.

And Luis, I think you've been asked this before, maybe a couple of quarters ago, just in terms of, you know, the size of TAM in terms of users. And some people do compare compares to the dating apps or companies. Maybe just kind of go back to that, give you an opportunity to have any updated thoughts and sort of like the long term penetration there.

Yeah, I mean, it's an excellent question. So for one, there's language learning as a whole. I mean, our best estimate, there's about 2 billion people in the world learning a foreign language. And collectively, they spend about $60 billion a year. So that's a standard answer. But it's a little more complicated than that, because in many countries, for example, in the US, 80% of our users were not learning a language before Duolingo. So it's kind of really hard to know exactly how many people we can get to. I mean, we're, you know, we're getting close to 100 million MAUs. We're like, you know, getting close there. I think we can get much farther than that. But, you know, it's just hard to know exactly how much because we're growing the market. Great.

Thanks, Luis. Thanks, Matt.

Thanks, Rolf. Next question comes from Eric Sheridan of Goldman Sachs.

Thanks so much for taking the question. Maybe come back to the element of what you see as some of the most critical investments in content. As we look out, not only to this year, but to the longer term, that could be elements of driving incremental user growth, time spent, engagement, and also monetization of the user base and how, I know you talked a little bit about it earlier, Luis, but how AI can sort of lower some of the friction to creating content versus some of it that might just be sort of elements of sort of building away from the AI initiatives as well. Thanks so much.

Yeah, excellent. Thank you, Eric. So let's see. In terms of content, the major things that we need to invest on in terms of content are, number one, more advanced content for some of our courses. So like we've said in the past, not all our courses cover to the same level of proficiency. I mean, for example, our course to learn French covers a lot more than our course to learn Italian. So we need to basically get all our courses to teach through pretty high levels of proficiency. So that's one place where we're adding content. In particular, we are interested in adding a lot more content to our English courses, our courses that teach English, because we know there's a large opportunity there of English learners. In general in the world, English learners If you take it just an average English learner they're usually more advanced than an average Spanish learners because just kind of you know worldwide they usually have some. Previous knowledge for for English English learners so so so there's a lot of content that we were we are adding. to our English courses, but to all of our courses to be more advanced. That's kind of one big chunk. Another big chunk is just making our content more interesting, and in particular, making it have kind of a narrative with all of our characters. So we've been working on adding stories and storylines for all of our characters and making them better and better over time. um so those are kind of two big initiatives in both of those cases ai can help us and we've been working hard to make it so that with ai we can generate it faster and cheaper um and uh you know the hope is that it's also of the same quality i would say for things like uh creating you know stories that are fun We can probably do it faster and cheaper. I don't think we're quite yet there of the same quality as very, you know, essentially Hollywood writers or something like that. We're getting there, but it's not quite there yet. So that's kind of what we're working on. And we do think that AI can help us really do much better there. Yeah, I think that's what I'll stop there. I think that's what I have to say.

Great. Thank you so much. Thank you, Ralph. Oh, Eric.

OK. So next question comes from Andrew Boone at JMP.

Thanks for taking my questions. I wanted to go back to one of the IPO disclosures and just talk about the 40% annual retention rate that you guys talked about back then. Can you talk about how that's trended? And then maybe how do we think about updating models today now that we're two years out from pretty significant cohorts? How do we think about maybe year two or year three?

Yeah. Yeah, it's a great question. So the way we think about retention in general is not in an end in itself. It's a piece of what we optimize for on the platform, which is lifetime value, which is a combination of pricing, packaging, and retention. So when you look at since the IPO, our platform LTV has increased a lot because we've shifted retention up on a blended basis because we have more annual subscribers and they have better retention. That's increased platform LTV. We've also have a brand new SKU, the family plan, which is a higher price and higher retention SKU. So that has also shifted platform LTV up. So on a platform LTV basis, we've increased that value. When you look at overall blended retention, it's about the same as it was when we went public. Some of the plans, like annual plan or monthly, some of those retentions have moved around. But when you blend it all together, it's basically the same as when we went public. And we think that's likely to be stable, at least throughout the rest of the year. So that's how we're thinking about retention as part of increasing platform LTV.

And then Luis, you talked about more of the advanced learner market earlier. What are the key product levers that you need to unlock there to really open up that market? And then just help us size that, right? Is that the majority of the 60 billion or is that a minority? Like, how do you think about the potential there? Thanks so much, guys.

Yeah, this is a great question. It is a huge opportunity. If you look at just the total language learning market, the majority of it is people learning English. And many of these people do want something more advanced than just beginner stuff. So this is why we're working a lot on adding a lot more content to our English courses. I think it's just a major opportunity and it's a weird thing because If you look at where our revenue comes from, the majority of our revenue comes from, for example, half our revenue roughly comes from the United States. These are usually not people learning English. They're learning kind of Spanish or French, et cetera. But if you look at the language learning market as a whole, it's highly skewed towards people learning English in non-English speaking countries. So you see them there in Asia, Latin America, Western Europe, et cetera, non-English speaking countries. We're kind of flipped but it's because in part because we're a digital product and you know We make our revenue kind of like how digital products make the revenue if you look at worse, you know Spotify or Netflix or stuff like that make the revenue. It's kind of similar to us We make it from the US and kind of English speaking countries, etc so so there's that but Because of the language learning market, we think there's a huge opportunity to make a lot of our money from different markets. And the unlocks that we need to do there, one, is we just need to have all of our English courses reach more advanced levels. We are working on that. And it'll be a few more months until we have that. You know, that by itself is not the only thing that will unlock more revenue from English learners. That's one thing that's needed. Another thing that's needed is truthfully in our product, we need to get better at whenever you come in with prior proficiency, we need to get better at placing you. That's another big unlock that we need to do. We just need to get better at that. And then after that, we need to get better at convincing, especially English learners, that we have this more advanced content. So we need to do that. And then after that, or around that same time, we also need to get to the point where in some of these countries we have not yet cracked how we're going to get them to pay. Because in many of these countries, There's just a lot less likelihood to pay, even if you reduce the price to something that makes sense in GDP per capita. They're always in the sense of like, well, I'm not going to pay if I don't have to. And our product is freemium, and many of them are just like, well, the free product is kind of good enough. So we need to unlock that. And we've seen a few companies do a better job than us. For example, Spotify has done a better job than us at that. So I think we just need to experiment on how to do that. So there's a number of things that need to happen. But once all of those things happen, which we are working on, we think this is a major opportunity for us. I mean, it's the majority of the market. Thank you.

And next, we have a question from Arvind Ramnani at Piper Sandler.

Thanks, Debbie. You know, I want to ask about this daily average users. You know, I mean, certainly it's been kind of tracking at a really healthy kind of pace. You know, how should we think like externally as like, you know, I mean, the higher the number is obviously better, but what's like a healthy number either in terms of growth rate or in terms of like total numbers?

I mean.

Go ahead, Luis. You want to take the first stab at it?

Well, that's a very hard question to answer. I mean, look, the way we see it is this is the main thing we are optimizing for. Because of our freemium model, the more daily active users we have, the more people we can convert to payers, et cetera. I mean, it's just generally good. And also, the more daily active users we have, the more data we have to run more experiments to be able to personalize things for you better, et cetera. So it's just a gift that keeps on giving when we have more daily active users. So for us, that's just the more we can grow it, the better. We're very happy with the fact that for the last eight quarters in a row, our growth has been either accelerating or remaining very high. Um, it's very hard to say how many more quarters we're going to have like this. And obviously we cannot have accelerating user growth forever that, that just, you know, at some points you run or humans. Um, so, but, but it is, you know, on our end, uh, I don't know, for me, it's very hard to say what's a healthy number. We just need to, we just need to continue growing and we've been.

Yeah. Yeah. I mean, many ways it's like money, right? Like more of it is good. It solves a lot of problems, but like, I'm just trying to figure out, like, is there like, like, is there a ratio in terms of like, you know, paid users was like, I mean, I'm sure you're going to some thought of like, what, what's a number where below it gets to below, below particular level, you're going to go get more aggressive in terms of like paying, paying for users or any of that.

Yeah, well, one way to approach this, Arvind, was before we went public, we had a lot of years of data. And user growth then was in the 25%, 30%-ish range. And that was really healthy. It got us to a nice scale. Now, I don't think that anyone on this call, Luis and I and our teams, would be super happy if next quarter that's where growth went to. But certainly, that was healthy pre-IPO. So at some point, that'll be a healthy growth rate in the future. Again, I don't think that's next quarter. But you can still produce an enormous amount of subscribers, revenue, bookings, and profit if your growth rate was in that level. And we think that level should be attainable or achievable for over the long term. So that's one way to kind of approach what's like healthy or not healthy. By the way, right now it feels healthy though. So just.

Yeah, of course. And then just on AI, you've given us a fair bit of commentary on AI and Duolingo Max is sort of the first sort of tangible kind of product. Is there anything you can provide To us, like in terms of like, where do you think like, you know, applications of, where you look to apply AI over the next like six to 12 months in terms of like a tangible product where you're able to actually charge for it.

I mean, in terms of what we're going to be charging for, the majority of our efforts are going to be on Duolingo Max. There are other places where we're applying AI within the company, and we've been. I mean, for example, we're making our content faster. We're making our content creation cheaper. We're doing a lot of things inside the company to do that. But in terms of things that we're going to sell to users, I think most of the effort is going to be on Duolingo Max.

Yeah, terrific. By the way, great start to your earnings call. I don't know if you can top that on the next one, but I thought bringing out Dio last earnings call was good, but you all outed yourself. Let's see if you're able to top that next earnings.

Thanks, Arvind.

I should say all of these celebrities, we do not pay celebrities for doing that. They use us.

Thanks, Arvind. Next question comes from Mark Mahaney at Evercore.

Okay. Thanks, Debbie. I had one or two for Matt and one or two for Luis. Matt, this really strong growth in MAUs, do you want to give us any color as to whether there's anything unusual in terms of where these MAUs are coming from? Any reason to think that the most recent surge in growth that you've seen in MAUs comes from markets that are more dependable or less dependable than what you've seen in the past? And the second finance question for you is... This is the first quarter in which your incremental margins were actually above your 30% to 35% long-term EBITDA margins. So the trend just seems pretty positive into the back half of the year. I don't know. At what point do you reconsider where your margins can go long-term? Or is it that you may want to be leaning more into investments? Do you feel like near-term you're over-earning and you want to lean more into investments as we go through the next 12 to 18 months? Yeah.

Thanks, Mark. Great, great questions, as always. In terms of the user growth, I'll say three things. So the first one is that we saw a great top of the funnel user growth and great kind of current user or active user retention and growth. So it's really across the funnel. We saw really strong growth and have for the past several quarters. The second thing is that it has been broad based geographically. So one of the things we look at is, you know, is it lopsided or is it basically all around the average? Obviously, around the average, things move. There's some that are going faster, some are going slower, but it's still broad based, meaning that, like, you know, the U.S. is growing really, really, really well. Western Europe is growing very fast. And then, you know, plenty in Asia is growing fast as well. So broad based geographically. And then the last thing we look at is, you know, how strong are these cohorts in terms of free to pay conversion? And we've seen great trends on free to pay conversion as well. So we feel not only is the user growth fast and impressive, but it's high quality and broad based. So we feel really good about that performance to date. And then we feel good about the forward on it as well. You know, on your second question, the incremental margin is the right one to think about it. And I'm glad you framed it in terms of the long-term margin. You know, we have not updated our view on long-term margin. We still think it's the same as it was when we went public, you know, 30 to 35% long-term margin. adjusted EBITDA margins. What we've shown this year is that there's power in this model. It's a very profitable model. And so we can get to that level of incremental profit slightly above this quarter. But I think, again, I'll just go back to our capital allocation strategy that I mentioned in the answer to another question. First and foremost, we're at the early stages of growing this business. It can be a very large business. And so we're going to continue to invest into the business. while also maintaining that profitability threshold that we talked about getting towards those long-term margins. But you should not read into this quarter as us redefining the long-term on that.

Okay. And then, Luis, two quick questions. Just any update on my favorite math subject, just what kind of traction you're seeing. And then your comment about... building out more content. Maybe think about like vertical language. You know, there's English and then there's legal English or there's English and then there's medical English. Like, is that something that you think about long-term? Like, you know, and I know there's different terminology. You can get very sophisticated, very complicated. I would also think highly profitable for any company that can solve, you know, learning medical English for somebody coming, making this up from in or from Guatemala, you know, whatever the example is. So is that a possibility, you know, kind of verticalizing and really getting to kind of, I don't know, whatever that is, professional English for different industries?

Yeah, so both great questions. I mean, in terms of math, we're very happy with the progress so far. It continues being the case that the app is growing and the growth looks a lot like early Duolingo. We're very happy with that. In terms of what we're doing and what we need to do, we know we need to add more content to it, just more subjects within math. And we're working hard on doing that. So you'll see us over the next few months, there will just be significantly more content. And that's the main thing. But the growth looks good. I'm very happy with it. In terms of kind of verticals of English, we know that this is a big business kind of things, like business English, medical English, et cetera. for now that is not something we're we're addressing you know in the future we may but right now we're just a lot more keyed in on really getting all of our english courses to get to teach pretty advanced english even though it may not be specialized but pretty advanced english at some point we may get into specialized english the one thing that i'll say about specialized english is the audience is start getting pretty small And we do much better with large audiences. I mean, if you think about it, we're an app business. It's rare to find apps where people pay a ton of money. Like a single person pays thousands of dollars to learn on an app. That's rare to find. And when the audiences are small, if you want to make a lot of money, you need to charge a lot. And so we're more in the business of getting very large audiences. And so that's why we haven't done it yet. But at some point, we'll probably investigate it.

Thanks, Luis. Thanks, Matt.

Thank you, Mark.

Thanks, Mark. And now we have a question from Justin Patterson of KeyBank.

Great. Thank you very much. I'll try to keep it short and big picture. I actually kind of want to bridge back to something both Andrew and Mario brought up in their questions. Time spent is obviously a great focal point in the app. On the one side, you've got younger generations that you know, have short attention span. On the other hand, you've got older consumers who also have limited time to spend on the app. So if you look at something like Duo Max, the premium here, how can you really kind of position this to unlock some of those pain points while also, Luis, being congruent with what you've told Andrew, finding that right price point to really succeed in the international market that have lower propensity to pay and just succeed on that midfitter, making this a global education platform. Thank you.

I mean, I think generally, you know, we're going to be testing a lot of the way we work with all of our products is we kind of test our way into things. I'm not entirely sure, you know, how exactly we're going to be, you know, addressing younger versus older audiences with Max, etc. I think we'll just test a lot and find a sweet spot where because of the price and because of the features that are being offered, you know, a pretty large number of people are buying Macs. But ultimately, I'll refer to just the fact that for all of these, it's hard for us to know in advance what things will resonate with which users, so this is why we test. Part of the reason is that our user base is just so broad, it's not just older versus younger i mean we have users in the lower economic or in the higher economic end of the spectrum we have users in every single country in the world etc so we just it's hard to think about in terms of for us a priority but but in the end we'll just um uh test our way to it that's that's what we'll find all right thank you and uh looks like we have a question from zach morrissey of wolf research great thanks debbie

I'm just curious on an update in terms of you called out paid influencers spend in Asia and LATAM last quarter kind of working well. Just curious for an update there. And if you see an opportunity to kind of translate that kind of strategy also to kind of core Western markets as well, if you're seeing kind of good returns.

Yeah, I mean, generally, I'll say for our marketing, per our shareholder letter this time, the vast majority of our marketing is geared towards just viral stuff that we don't even pay for. In some cases, particularly in Asia and Latam, we have found that working with paid influencers actually makes a lot of sense. And we've gotten really good results with that. we have not gotten as good results in places like the US for influencers. And in part, that's because influencers in the US are just very expensive and we don't see the returns. I mean, for similar types of video views or something like that in the US versus in a place like Brazil or something, the difference in price that you have to pay an influencer, I don't know off the top of my head, but it's something like 10x. And you just don't get the 10x returns. It's just much more expensive to pay Beyonce to do something versus to pay whatever the equivalent in Brazil is of a Beyonce. It's just much cheaper. So we have not cracked that, and I don't know if we ever will. For now, in markets like the U.S., Our marketing is just much more, you know, our owl doing unhinged stuff on TikTok or on YouTube shorts and stuff like that. And that has worked really well.

Got it. That makes sense. And then just one on, you know, Gen AI, right? I think you kind of talked about a healthy product roadmap for Max. Do you see an opportunity for, you know, in-app purchases products from Gen AI or is the primary focus right now just driving kind of a premium tier on Max and maybe in-app purchases is something down the road that you can kind of layer in over time?

It's the latter. I mean, there's definitely an opportunity for in-app purchases for almost everything we do. But, you know, historically, our subscription business is just so good that that's where we're putting our effort in. And at some point, you know, once we've really nailed what Macs will be like, we may start selling some of these features a la carte or we may start having kind of power-ups or something like that. But for now, the effort is going into the higher tiers description.

Makes sense. Great. Thank you.

Thanks, Zach. All right. So I'm not showing any other further questions. So I'll turn it back to Luis to wrap up.

Thank you, Debbie. I'd just like to thank everyone for joining us, and we look forward to speaking to you in November. If you haven't watched it, watch the Barbie movie. We're there. And have a great evening.