Good evening, everyone. And welcome to Duolingo’s Second Quarter 2024 Earnings Webcast. I’m Debbie Belevan, Head of Investor Relations. Today, after market close, we released this quarter shareholder letter, a copy of which you can find on our IR website at investors.duolingo.com. On today’s call, we have Luis von Ahn, our Co-Founder and CEO; and Matt Skaruppa, our CFO. They’ll 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, this event is being recorded and all attendees are in listen-only mode. Just a reminder, 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 with that, I will turn it over to Luis.

Thank you, Debbie, and welcome, everyone. I’m excited to report that we delivered another strong quarter, with daily active users growing 59% and revenue growing 41% year-over-year. We had a record adjusted EBITDA margin of 27% and this marked our fifth consecutive quarter of being net income positive. We also achieved two milestones, surpassing 100 million monthly active users and reaching 8 million paying subscribers. This is particularly impressive considering that only three years ago, at the time of our IPO, we had 38 million MAUs and just under 2 million subscribers. Our top-of-funnel growth remains strong, driven both by new users and those returning to the app after having been away for over a month, and that’s across both more and less penetrated markets. Our free, fun, and effective app gets better over time through continuous testing and iteration, and that’s why in Q2, our DAU growth accelerated and we reached a record-high 33% DAU to MAU ratio. And not only did our business perform well, but we’re also making substantial progress on our long-term growth initiatives. This year, our monetization priorities are optimizing both our family plan and our tiered subscription plans so that we can offer learners more choices at various price points and increase LTV. We continue to see excellent growth in our family plan. We’ve rolled out improvements to increase engagement between family members and we’re also helping existing individual subscribers discover and convert to family plans more often. As a result of these efforts, the family plan now makes up about 20% of our subscribers. At the end of Q2, we began to see the impact of Duolingo Max, our highest-priced tier with AI-enabled features. The rollout of Max has progressed so that, as of now, it’s available in five courses in 27 countries, covering about 15% of our DAUs. During Q3, you will see us expand the number of countries Max is available in, and by year-end, we expect it to be available in most countries on both Android and iOS. This will set us up to see the impact of Max more fully in 2025. We will also be introducing new Max features that we believe will resonate with learners. This includes AI-powered immersive conversational practice, one of the most requested features over the last 10 years. Join us at Duocon on September 24th for the unveiling of this new magical experience. I want to conclude by talking about our initiative to better serve advanced English learners, which will allow us to more effectively reach the approximately 1.6 billion English learners who aren’t on Duolingo today. We now have more advanced content in all 20 of our English courses, as well as a standalone English course for intermediate to advanced learners. We have exciting plans for the addition of DuoRadio and Stories to the more advanced levels of these courses. We’re also improving how we place learners with prior proficiency, so that they feel adequately challenged, but without hurting their engagement. As a reminder, our English learners’ initiative is a multiyear effort that we expect to monetize over the medium to long-term. As you can see, we believe in investing in innovation to drive both current and long-term growth. We have very ambitious goals for our business and believe that Duolingo Max and our more advanced English content give us more options than ever before to achieve them. Of course, this path will not be linear, but we’re excited by the huge growth opportunity we see ahead. And with that, I’ll turn it over to Matt.

Thanks, Luis. I’ll provide some additional color on our Q2 results and then update our guidance for the remainder of the year. To emphasize what Luis said in his remarks, Q2 was a strong quarter for us. We reaccelerated DAU and MAU growth to 59% and 40%, respectively. We continue to see healthy top-of-funnel growth and we saw particular strength in resurrected users—users who come back to the platform after more than 30 days away—thanks to a seasonal feature we launched that focused on reminding them to do their lessons. In Q2, bookings grew 38% year-over-year, revenue grew 41% year-over-year and we posted our highest quarterly adjusted EBITDA and adjusted EBITDA margin. And as Luis mentioned, this is our fifth consecutive quarter being net income positive. All around, it was a strong quarter. Going forward into the second half of the year, we feel good about the business, even as we lap the incredible strength we saw in the back half of last year. For the full year, we are raising our guidance so that at the midpoint, we are guiding to year-over-year bookings and revenue growth of about 32.5% and 38.5%, respectively. Our guidance implies a year-over-year growth rate for the second half of the year of about 27%. Note that if FX rates were constant year-over-year, our Q3 bookings growth rate would be about 2 points higher and our full year bookings growth rate would be about 1.9. To put our topline growth into context, we grew bookings at about 40% year-over-year in the first half of this year, which is roughly the same rate as we grew bookings in the first half of 2023. In the second half of 2023 though, our year-over-year bookings growth accelerated materially up to about 50%. A meaningful part of that acceleration came from an extraordinary set of tailwinds. Our signature Duolingo was featured in the Barbie movie, providing an incredible brand boost and in many ways that was indicative of our commanding presence in the zeitgeist last year. It also helped our summer marketing campaign deliver home-run user growth. That user growth was coupled with some one-time improvements in monetization, especially in Q3, and more favorable exchange rates than today. This year we feel good about our user growth and monetization, but we don’t expect to see the same one-off accelerating tailwinds. Lapping last year’s extraordinary growth is why we expect DAU growth to decelerate to about 50% in the second half, and why we expect bookings and revenue to decelerate as well. Moving down the income statement, we expect gross margin will go down slightly in the back half of the year, as our amortization and AI costs increase with the rollout of Max. We continue to feel confident about our ability to drive increasing profitability and are raising our 2024 adjusted EBITDA margin guidance to 24.5% at the midpoint, which is a full 7 points higher than 2023, and is an incremental margin of 42%. For Q3, our adjusted EBITDA guide is about $41 million or roughly 80% higher than the same quarter last year. As a reminder, our profitability typically varies a bit from quarter to quarter throughout the year. For Q3, we expect to see about 5 points of quarter-over-quarter deleverage in the margin and are guiding to 22.0% at the midpoint. This is mainly driven by increased hiring and R&D, as we add a significant portion of our new engineering, product, and design hires in Q3. We’ll also see seasonally higher sales and marketing spend this quarter. G&A should remain relatively flat as a percentage of revenue. For Q4, we expect to see slight leverage across all OpEx categories, resulting in about 1 point of quarter-over-quarter margin expansion compared to Q3. Note that overall, we shifted about $3 million of expenses from Q2 into the second half of the year. Finally, we ended the quarter with approximately 49.4 million fully diluted shares outstanding, using the quarter-end closing price. In 2024, we expect to end the year with about 1% to 1.5% net dilution from equity issued to employees, which is similar to the dilution we had last year. And with that, I’ll turn it back to Luis.

Thank you, Matt. As you can tell, I’m excited about the road ahead and confident in our ability to achieve our long-term goals. And now we would be happy to take your questions. I’ll turn it back to Debbie to manage the queue.

Okay. Thanks, Luis. And as I mentioned earlier, if you have a question, just use the raise hand feature. And our first question comes from Andrew Boone at JMP Securities.

Thanks so much for taking my questions. Luis, you just defined the top-of-funnel as being healthy for new and returning users. Some third-party data sources suggest that downloads have slowed. Maybe this is exactly what the comp that Matt’s talking about. But is there anything you guys can help us understand as we think about new user growth and the opportunity to bring more users into the Duolingo ecosystem?

Yeah. Thanks for the question, Andrew. So, first of all, we don’t comment on third-party data; if you want to watch it or see it, you do you. But, in general, what we’d like to say is we have users in every single country in the world. Some countries are more mature than others, because we’ve been there for much longer. So the way we think about it internally is more top-of-funnel, which is brand new users, usually downloads and they’re brand new users. But we also look at returning users, people who haven’t been around the app for a while, typically longer than 30 days. In countries that are more penetrated, more mature, the number of people that are coming back after a long period of inactivity is much higher than the brand new users, just because we’ve touched a good fraction of the country. So generally, this is what we’re looking at and we feel pretty good about top-of-funnel because of that. It’s very rare to see people completely stop using Duolingo. What they do is they stop and then they come back six months later or they come back a year later and so that top-of-funnel feels pretty good.

So to follow up on that question, is there a way to think about cumulative users for Duolingo, right? If we think about the 2 billion people that are actively learning a language, how do we think about the number of learners that you guys have touched within that opportunity? Thanks so much.

It’s a good question. That’s not something we really track. So I don’t really know it off the top of my head. Yeah, it’s not something we really track.

Okay. Thanks, Andrew. And the next question comes from Bryan Smilek of JPMorgan.

Great. Thanks for taking the questions. So results were quite strong across DAU growth acceleration, engagement at record highs and social media impressions almost doubling year-on-year. So can you just talk about the health of the product, refresh cycle into the back half, and just overall social media strategies as you count stuff like Barbie from last year?

Yeah. I mean, product-wise, we’re very excited about a lot of the stuff that we’re about to either already testing or putting out. There are a lot of new social features. I mean, we talked about in our shareholder letter, we talked about the Friend Streak, which is basically a streak that you keep with your friends. So our regular streak is just a streak with Duolingo. Now, you and a friend, it’s up to five friends, but you and a friend can have a streak, and you both have to use the app every day. If one of you doesn’t do it, you lose your Friend Streak. So it gives you a really good incentive to not disappoint your friends. So that’s an example of a feature; there are a lot of great social features that are coming out. We have a lot of monetization features. We’re also expanding a lot of our teaching features. So I feel pretty good about that, and because of that, we expect the growth to continue being strong. As Matt said, we are lapping a very strong year. So, we’re expecting that for a while our user growth is going to be about 50% and that’s kind of what we’re looking at. Now, in terms of social media impressions, we measure our social media impressions in billions. And this, by the way, to remind you, this is organic. We’re not paying for these impressions. So we’re getting billions of social impressions and we expect that to continue.

Awesome. And then I guess just to follow up on Max, available to 15% of DAUs right now. Could you just share some signals around monetization, when should we expect Max to show up in the P&L understanding it’s a multiyear endeavor? And I guess, like, is there anything that could curb the pace of rollout as we enter the back half?

Yeah. So, as you said, we’ve rolled out Max. It’s now in 27 countries, five courses, and that’s available. That’s about 15% of our DAUs. Our expectation is that we’re going to continue rolling it out over to—through the rest of the year. And by the end of the year, the majority of our users will have access to it, at least. And the reason for that is because we feel good about it. We’re giving it to more and more users because we feel good about it. We’re also not only rolling it out; we’re going to be adding new features to it. The one that I’m most excited about is actual conversational practice with one of our characters, Lily. If you don’t know which one that is, that’s the purple-haired girl that is not impressed at all about anything you do. So you’re going to be able to talk to her and it’s a really awesome experience. So our sense is that, our users are going to love it. Of course, it’s hard to know exactly without a feature that we haven’t really put out, but we’re very excited about it. And to reiterate what you said also, we are still iterating on Super Duolingo, which is something that we put out probably seven years ago. So we’re going to be iterating on Max for many years. It’s just by the end of the year, it’ll be at least mostly rolled out.

Awesome. Thank you for the color.

Thanks, Bryan. Next question comes from Ralph Schackart at William Blair.

Good afternoon. Thanks for taking the question. Luis, just curious on the Max product, as you think about that opportunity there as sort of an innovation platform, is that sort of an opportunity for you to develop and test new products and eventually sort of, I guess, see if those could roll out to some of the lower tiers over time? So the first question. The second question, I know you get this question from time to time, so I guess I’ll ask it. But often you’re asked, what percent of your paying subs do you see as sort of an opportunity? I think maybe last time you were asked that, maybe I’m wrong, is pre-Max or maybe before Max was scaled. But now that you have Max sort of in the fold, just kind of curious if you sort of give us an update on that as well? Thank you.

Okay. So, your first question is if Max is a testing ground for features that maybe we can then put in different tiers. That’s for sure true. Basically, the way we’re seeing it is we are committed to the three-tier strategy. We have free, we have Super, we have Max. We’re going to see where the features belong. Right now, a lot of the AI-based features are in Max because it costs us money to serve those, so they’re in the highest tier. Over time, we expect that the price of some of those AI features will come down, at least for the cost of them for us will come down, because we’re expecting that the price, that sort of the cost of LLMs is going down. At that time, we may make a decision that some of those features belong in Super or even the free tier. But it’s hard to say exactly what will happen. My sense is that if you look at it over time, what’s going to be happening is that the heaviest features on AI, the ones that use most calls to a large language model will probably remain in Max. And then anything that we can either cache or just doesn’t require that many calls will put in lower tiers. That’s probably what’s going to happen. Now, in terms of penetration, I think your question is just how high can penetration get either for Max or for paying—either for Max or any paying subscriber, like either Super or Max. We don’t know. I mean, in terms of penetration, we’re—overall paying subscribers, we’re—the latest number, we’re above 8%. We seem to be increasing about 1 percentage point every year. That’s what’s happened historically. We expect that to continue happening, but I don’t know the exact numbers. I mean, and it’s not clear to us where it’ll end. I mean, I think it’ll be higher than what it is now. But I don’t know if some people ask, is it 15%? Is it 30%? We just don’t know.

Okay. Great. Thanks, Luis.

Okay. Next up, we have Curt Nagle from BofA.

Great. Thanks for taking the question. Yeah. First, maybe on the family plan, right, progressing nicely up a bit since I think last time we spoke. You have a number of initiatives in the works to increase adoption. I guess just how fulsome are they now in terms of the actual product and where do you think we could see rates of total subs go by year-end?

Yeah. We’re—just to remind people of the family plan. So the family plan was something that we built a few years ago and didn’t touch. We just built it, left it there and it grew by itself, which was awesome. Once we realized it was growing so much by itself, we put a team behind it. That team has been there not that long. It’s a few months. So they’re kind of really getting their stride. But already they’ve made a number of changes and we’re starting—we’ve started seeing more and more penetration of the family plan. At this point, about 20% of our paying subscribers are paying for the family plan. And we expect that number to continue growing. Some people are going to ask us a similar question, the previous question, which is how high can that get? Again, same answer. I don’t know how high it can get, but we do expect it to get higher than 20%. And it’s because we’re just adding a lot more features. For example, things where families can engage with each other a lot more. And not only are we adding features, we’re also making the family plan a lot more well-known among our subscribers. Many of them don’t even know that there’s a family plan. So we’re just putting it in front of the right people more often. And so what I can say is that by the end of the year, we expect the number, the fraction of subscribers to be on the family plan to be higher than what it is now. But I just don’t know how much higher.

Right. Makes sense. And then just going back to some points on top-of-funnel, in terms of existing or last—however you want to define it, you’re just coming back. What do conversion rates look like in terms of paying users for those users? Is it higher than average?

Yeah. So…

I think Matt knows that one better than I do.

Yeah. When it comes to the actual kind of free trial conversion rates, they convert relatively similarly. The new users and the returning users, they start free trials at slightly different rates. And so we actually see a nice opportunity to continue to drive those rates higher for both types of users over time. So there’s good opportunity there for both.

Right. Thanks so much.

I should say, just one last color. Historically, we spent a lot more time optimizing new users than resurrected users, but we’re starting to work more on resurrected users because it just makes sense to do that.

Make sense.

Next question comes from Alex Sklar of Raymond James.

Great. Thank you. Luis or Matt, just following up on Bryan’s earlier question on that new Max availability disclosure. Can you talk about how that availability looks as a percentage of your existing paid subscriber base? And I’m curious if you’re seeing any notable difference in terms of Max adoption from free users versus Max adoption from your Super base today?

Matt, I think it’s pretty similar to paying subscribers. I think it’s pretty similar to the users—probably 15% in terms of availability.

Yeah. I think, definitely it will over the next little bit, it will converge. So they’ll be close over the next couple of quarters for sure.

And now in terms of, we—there’s appetite for Max from both free users and also from current paying subscribers. So we’re working on both, putting the offers out for the free users, but also cross-grading from the paying subscribers, and we’re seeing uptake from both.

Okay, I appreciate that color. And then Matt, maybe just one follow-up for you on the subscription bookings continuing to trend really nicely and now making up a bigger piece of total bookings. I was just curious on the non-subscription bookings, so that other three buckets, can you comment on what you saw in trends in the bookings there and anything to flag for the outlook?

Yeah. No. You’re absolutely right to start with the fact that we are a subscription business and you’re going to continue to see our subscription business grow faster than the other lines. And so over time, what you just—the trend you just identified will continue, and that’s by design. We think that the subscription product is a better user experience and it’s a better economic vehicle for us. So I think you’ll expect those trends to continue. Each of the other lines of business makes up less than 10% of revenue. So, ads is 8%, IAP is 6%, DET is 6%, something like this. So they have trends in them. Ads, for example, has grown a little slower so far this year. So has IAP. Both of those have been—we don’t have a ton of resources working to grow those. I think in the back half of the year we’ll probably devote a little bit more resources to both as an IAP, so they might grow a little bit faster, but still slower than subscriptions. DET is a—as you know, is our testing product. That business has been growing nicely. I think in the back half of the year, we expect it to grow a little bit slower than it did last year, because there’s kind of an industry-wide visa issue going on, which has slowed down some of the test-taking growth. But again, all of these movements are pretty small and on the margin. Again, we’re going to be focused on driving subscription growth higher and like you said, it’s been growing really nicely.

Perfect. Thank you both.

Okay. The next question comes from Shweta Khajuria at Wolfe Research.

Hello. Thanks for taking my question. Just wanted to ask about the resilience in the softening macro environment. You have the subscription piece, you have the advertising piece, but just remind us what you’ve seen in the past as perhaps maybe there’s increasing concern on consumer spend and the macro environment right now? Thank you.

Yeah. That’s a good question. I mean, we’re seeing—what you’re seeing in terms of the external world. There’s a lot of uncertainty in the external world. When we look at our numbers, we’re not seeing anything. I mean, our consumers don’t seem to be reacting to anything. There’s nothing worrying. In the past, we’ve had similar situations where there’s uncertainty in the world and when we look at our consumers, we just can’t see anything. And we don’t really know why that is. I’m not going to say that we are recession proof. We just don’t know because we’ve never really gone through a recession. But what I will say is that because we have such a good free tier, what’s happening is that people who—for whom $6 a month is a lot of money, they’re just not paying us, because they’re just using the free tier. So whenever there’s a recession and the consumer is maybe cutting down, et cetera, we just don’t see it very much because the people who are paying us, for them, $6 a month is usually just not a lot. So we just haven’t seen anything like that.

Okay. Thanks a lot.

The next question comes from Ryan MacDonald at Needham.

Thanks for taking my questions and congrats on this quarter. Luis, I wanted to talk about in the shareholder letter the international efforts. And you mentioned the marketing efforts you’ve made in Japan as sort of impressive returns you’ve been generating from those investments and sort of that being a playbook for international expansion. Can you just, one, remind us when you started investing in Japan to sort of give a sense of what the payback period on those returns, and two, how quickly do you expect to start replicating that playbook in Japan into the other, I think, it was about 20 targeted countries or languages for that advanced English learner? Thanks.

Yeah. So, what we mentioned in the shareholder letter is that we now have a playbook for international marketing that we feel pretty good about in terms of penetrating new markets and Japan was an example. We’ve actually done that playbook in a number of other markets and it’s worked very well. I mean, these are markets like Germany and France, where the idea—and it’s a pretty similar playbook. The idea is to start at the beginning with a little bit of performance marketing to kickstart a little more growth than what was there before. Then you usually hire one or maybe more than one country marketing manager that starts making our usual social content, but adapted to that country. And then we start a TikTok account and an Instagram account, et cetera, adapted to that country, and it has worked every single time. Basically, if you look at our TikTok accounts across all different countries, they’re all very popular. The Japan one’s popular. The France one is popular. The Germany one is popular. All the different ones. The Brazil one is popular. So, and this works pretty well. It takes—to give you context, it takes about a year to get our act together in a country. Japan, we probably started investing more heavily probably about two and a half years ago, give or take, but we’ve been seeing returns in Japan for a little while now. So my sense is for a given country, it takes us about a year to get our act together.

That’s helpful. And then as you think, maybe as a follow-up to that, as you think about sort of the targeted sort of 20 countries or languages for the advanced English learner content and trying to bring those learners onto the platform, maybe how many of those countries today do you feel like you start to get your act together, so to speak? And just to give us a sense of where you are on that progression of that investment? Thanks.

Yeah. So, just to clarify something. So we have 20 English courses. What that means is that we have 20 different base languages from which you can learn English. So a base language could be Spanish, that is Spanish speakers learning English. And other ones, Chinese; Chinese speakers are learning English, et cetera. So it’s not exactly countries because, for example, Spanish covers Spain and all of Latin America, except for Brazil. So that’s—these courses are based on the language that people are speaking when we have 20 of those. And in addition to those 20, we have another course, which is monolingual. That means learning English from English. That’s to cover all other countries or the ones that don’t speak these 20 languages. And so, basically we’ll be able—with this, I think we’ll be able to cover pretty much the whole world and we’ll be able to penetrate them. And now in terms of the different geographies, there are a number of them that we’re just much less penetrated than we’re currently. I mean, typically we’re pretty highly penetrated in the U.S., maybe even Western Europe. But then outside of that, we’re just at the beginning stages. This includes countries in Asia. Even Japan, we’re just not all that penetrated in Japan, Korea, et cetera. Yeah.

Excellent. Appreciate the clarification. Thanks.

Okay. Next question comes from Arvind Ramnani at Piper Sandler.

Thanks for taking my question. Yeah. I guess, just a couple of quick questions in terms of the stat that you shared, 20% of your users have been using, they have a 365-day streak…

Or longer.

…longer. Can you provide some historical context? Like how has that moved up or like, what should we look at that? And I mean, what do we make out of that number? I mean, it’s a nice, interesting stat. You chose to put it on Page 1. Can you just share, like, what gets the excitement and historical perspective?

Yeah. It’s a very exciting stat for us, because it just puts it in perspective. That means 20%, that’s like—it’s about 7 million daily active users have a streak longer than 365, meaning they have used Duolingo every single day for a year or longer. That—we find that that’s pretty impressive. I mean, there are 7 million of these people that have kind of not a day for at least a year or longer. That number keeps growing and it’s growing pretty fast. I mean, I think we put out a similar number, the same stat, maybe a year and a half ago or so, and it was like 3 million back then.

Yeah.

So that number just keeps going up pretty fast and it’s because—that just gives you an idea of how sticky the product is.

Yeah. And I guess, like, can you share, not necessarily a specific number here, but like, these people are sticky. Do they end up having a disproportionate number of paid users or unpaid users compared to the other 80%?

Yeah. I mean, in general, the more you use Duolingo, the more likely you are to subscribe. That’s, of course, not true for everybody. But in general, if you have a long streak, it is more likely that you’re a subscriber. And that makes sense; you get more use out of it and at some point you want to either support us or turn off the ads or something like that.

Okay. Perfect. And then just one last question. I mean, certainly very impressive metrics through the print and you raised guidance for the year. I get it. But it feels like a lot of the guide was raised in line with the Tokyo performance. So, I mean, just given the momentum, like, should we look at this kind of raised guidance as like a little bit conservative or how should we look at it? Because it feels like there’s really good momentum in the business?

Yeah. There is good momentum in the business. We feel good about it. But I just have to remind you in terms of what good momentum looks like. We’re guiding this year to more than 30% growth in bookings. But we’re lapping a year that had 45% growth on bookings, which in turn was lapping another year that had 45% growth on bookings. So we just—it’s just growing really fast and so that’s that. We feel pretty strong about that. Now, because we’re lapping such a strong year, in particular, H2 of last year was quite strong. So, we put out the guidance that we saw where slightly raced it, but we feel we feel good about it.

Yeah. Then this last question, I mean, it feels like your DAU growth is again back to sort of defying gravity, right? Like it’s just back to this, like, kind of acceleration. And like, I know that in the past couple of quarters, you’ve got it—late last year, you were talking a little bit conservatively, but you’re back to this, like acceleration in DAU growth. And like, I mean, how sustainable is this, right? Like, is this like 50 by 60 is like a number we should get kind of comfortable with over the next couple of years? I mean, it just feels like it’s constantly defying gravity?

We feel good about our DAU growth. I will say, in Matt’s prepared remarks, our belief is that this is going to for a while stay at around 50%, not quite 60%, but it’s for a while. So that’s our belief. And which it’s exponential growth, 50% year-over-year and I—that’s kind of how we think it’s going to be.

Perfect. Thank you.

Thanks, Arvind. Next question comes from Aaron Kessler at Seaport.

Great. Can you hear me?

Yeah.

Great. Maybe just on other verticals adoption, I don’t think you talked much about that. Just anything to highlight on some of the other verticals, math, et cetera. And then I think there’s still probably a little bit of concern in the market just around potentially AI replacing maybe the necessity to learn a foreign language. Just be good to get your thoughts on that, Luis. Just maybe your thoughts on potential AI replacing the need for a foreign language learning longer term.

Sure. Okay. So, first question was about other verticals; in particular, you can now learn math and music on Duolingo. We feel very good about the growth. They’re still small compared to languages and they’re going to be small for a little while. They’re only on iOS. We’re about to put them on Android in the next couple of months. You’re going to see them on Android and that’ll make them grow more. But we feel pretty good about the growth. In terms of what we’re working on each one of them. In both cases, really, we’re working on adding a lot more content. With math, we probably have an order of magnitude less content than we would have for like French or Spanish. So we just need to add more content. And same with music. And so we’re working on adding more content and also on making them more engaging. And I feel pretty good about the results of that. In terms of AI, this is not—people say this, but this is not something we’re particularly worried about for a number of reasons. For one, language translation and also automatic immediate voice-to-voice language translation has been really good for like 10 years. I mean, Google Translate, the app is on my phone, has been there for a while and it’s been really good. The main use for that—and it works really well—is if you kind of don’t really care about learning any French and you go to France and you find somebody who doesn’t speak English, you can use the app and you can communicate really well. But that has been true for like 10 years. It turns out most of our users are not that user. Most of our users are either people who are hobbyists, because they’re like, just they’re interested in Swedish or whatever. And we just don’t think that AI is going to replace that. Similarly, to—people are learning chess and AI has been good at chess for 20 years. So hobbyists is one big group. And then the other big group is people who actually want to learn the language for either work purposes. These are particularly English learners. That’s also not a—you’re not going to go work at a company in English and have to carry your phone or have to use a some sort of headphone device. So it’s just not something we’re particularly worried about and we just haven’t seen that be any kind of problem for our users.

Great. Thank you.

Okay. Next question comes from Eric Sheridan at Goldman Sachs.

Thanks for taking the question. Maybe I can follow up on that point. We’d love to go a little bit deeper on how you’re thinking about an incremental dollar invested in content and how the return on that content investment can continue to drive longer and longer engagement sessions with folks and continue to extend some of the streak dynamics that you’ve called out and sort of maybe just tease out a little bit of how you think about return on that spend or duration of potential for that spend in the broader portfolio. That’d be number one. And then number two, I know I always ask Matt this, but just incremental margins, updated view, continues to be an area where you’re surprising more of the upside as the business continues to grow and mature, but just updated views on incremental margins and investing back in the business versus letting some of that incremental margin fall to the bottomline? Thanks so much.

I can take the first part, Matt, with the content. I mean, generally, we are—we’ve been adding content to the app or ever since we launched the app is, 12 years ago or whatever. Over the last year and a half, we’ve seen an amazing acceleration in terms of the speed of content that we can add. A lot of it has to do with AI. Not entirely. Some of it is just better processes from ourselves. But a lot of it has to do with AI and we’re really excited about that. And it’s not just that it’s cheaper to add the content. That’s interesting and that’s good. It helps our margins and I’m sure Matt loves that. But what I’m more excited about is that, nowadays, we can create certain pieces of content so fast that these are things that just we couldn’t do before. A good example is, a few years ago, a feature was proposed to me that, it was a feature where you could like listen to these kind of podcast-type episodes inside the app and I was told that it was going to take five years to create the content for this and I immediately shut down that feature. I said, no, no, no, we’re not going to make that. I don’t want to spend five years creating content for anything. Today, we ended up building a pretty similar feature and it turns out we can create all of that content in a matter of months. And because we can create all that content in a matter of months, I greenlit that feature because I’m like, yeah, sure, why not spend a couple of months doing the content and do that. So I’m very excited about the fact that this just allows us to create a lot more engaging content and things that before they—they’re not that they were impossible to create, but they were prohibitive to the point where I was just not allowing it. So that’s one of the main things that we’re excited about with content.

Yeah. And then on the incremental margin point, Eric, again, we are definitely proud of the fact that, when we—in 2022, our adjusted EBITDA margin was 4 points and now we’re guiding it the full year to 24.5%. So, 20 extra points in two years of margin is great and requires higher incremental margins. A minor point on Q2, we had about a 52% incremental margin. If you were to adjust for the time shift that I mentioned in my prepared remarks, it’s probably 45% incremental margin, so lower; still higher than our long-term target. If you look at the full year guide, obviously we’re projecting the incremental margins to trend down a little bit in the back half of the year. But your point notwithstanding, we think that we can do both. We—the business has scaled really nicely. It’s got high gross margins, so we can achieve our capital allocation strategy, which is first and foremost, investing in R&D in the business to drive organic growth in that flywheel and drop a good portion down to the bottom line. So I don’t see it changing in the near term or any updating of that long-term target, but we’re happy with it.

Thanks, Eric. Our next caller—our next questioner is Chris Kuntarich from UBS. He cannot get on Zoom right now, so he asked me to read his question. Two-part question. First, housekeeping on the 50% DAU growth comment. Was that previously low to mid-50s? And his second question is, can you talk a bit about what you’ve accomplished over the last 12 months from a product monetization perspective as it relates to the English learning opportunity and how we should think about your focus areas on the English learning opportunity over the next 12 months?

Okay. The first is the clarification. I mean, I think it’s just—what we are saying is this quarter, Q2, we saw 59% DAU growth year-over-year. What we’re saying is that we expect that to be around 50% for a good period of time. That’s what we’re saying. In terms of what we’ve done for English learners, some number of things. I mean, the first thing is we’ve added the more advanced content. That content, it took us a while to add it, but at this point, the content is there, and so—that’s great. We have intermediate to advanced content for English learners. That was needed in order to attack the English learners. We’ve also worked on placing the English learners into the appropriate place in the course. So these are people who have prior proficiency. The thing about English learners is that most of them have some amount of prior proficiency, and so we’ve worked on placing them in the right place. And then, the next step is once we’re really happy with this, we’re going to start marketing to let the world know that Duolingo is good for advanced English, because most people just don’t know that. And once that’s there, my sense is that we’re going to be able to do a lot of things to monetize these users. But the first thing is to get these users to come in, which they’re not there yet.

Okay. Next question comes from Ross Sandler at Barclays.

Great. I had a question to follow up on the topic of AI, how you guys are using it in the app. So Luis, the foundation models are getting much more performant, faster, and more multimodal. So how’s that informing your product roadmap? And how do you see the interaction layer of your consumers engaging with either, Lily or whatever you guys come up with down the road in a more multimodal context using these AI models?

Yeah. We’re very excited about AI in general. I mean, the two big uses are creating content that is kind of pre-made and then it’s served to the users. That’s kind of what I was talking about in the previous question where we’re able to create experiences like listening to a 2-minute mini podcast inside the language that you’re learning. All of that is—a lot of it is created with AI and we’re very excited with that. And then the other big part is kind of live interaction where you’re going to be talking to a character, et cetera. We’re going to be putting a lot into that. Certainly, the multimodal models are helping on my like GPT-4o, because we’re kind of be—we are going to be doing voice-to-voice. And so that helps a lot. And so we’re really happy. I mean, if you play around with this feature of practicing conversation with Lily, it’s pretty magical actually. The other amazing thing about it is she has a memory; she remembers. So whenever she calls you or you call her, she tells you, oh, it’s you again. And then she may ask you about whatever it is that you were talking about last time. She’ll make fun of you for having forgotten something. So it’s pretty magical actually.

Okay. Great. Next question comes from Mark Mahaney at Evercore.

Super. Two questions. One, I love the Japan data point and what I want to try to figure out is I think in the U.S. you have kind of high single-digit percent penetration of people who are learning languages and interested in learning languages. That’s their own work. Are there any international markets and maybe it’s Japan that come anywhere close to that where you think that you’ve got penetration that’s kind of high single digits or even double digits of the language, potential language learners in that market?

Yeah. The U.S. is a relatively highly penetrated country for us. There are others that are probably similar to the U.S. My sense is Western Europe is close to the U.S. Actually, the U.K. is about the same as the U.S. But Western Europe is close to the U.S. Japan is not there yet. Japan is much less penetrated than the U.S. And if you look at other Asian countries, I mean, for example, China is about a tenth of the penetration as the U.S. So, generally U.S. and Western Europe are the more penetrated markets.

Okay. And then I want to follow up one more time. I know both Eric and Ross asked you this question. I’m really intrigued by the deployment of AI by content companies, by app companies. And investors in the street have so focused on the derivatives in the chip sector and in the app sector. I mean, sorry, the chip sector and the infrastructure sector, but not in the app sector. And there are companies, I mean, people so focused on how AI could disrupt Duolingo and there’s not enough focus on how you’re using AI and how you could use it in the future to create just dramatically more engaging content. Like it’s not like you haven’t been involved with AI for a long period of time. So, just talk about like when you think about how Gen AI tools can really change Duolingo and make it just dramatically more engaging, like, as you peer through that looking glass, like how much visibility do you think you have into how much more, how much better the product could be over the next couple of years? You think like you’ve got a quarter visibility and then there’s a whole bunch of new hallways that could open up. So just riff on that a little bit, Luis.

I’m generally extremely excited about what AI could do for us. Of course, the easiest to see is conversational practice, just because these are large language models—they’re supposed to have the LS language there. This is going to be really good for conversational practice. And I really think that’s actually going to be a lot more impactful than people think, because you really are going to be able to practice with somebody that gets down to your level of the language and she has a memory. And the other really nice thing about that is, it’s much better than practicing with a human in that. If you’re practicing with a human, you’re usually shy in the language that you’re learning. I mean, if you ever learn a language, it’s pretty hard unless you’re an extreme extrovert. It’s pretty hard to get your few words out in front of another person. But in front of an AI, you have no problem because you kind of don’t care. So we’re very excited about that. But I think there are going to be many other things. Another thing that I’m pretty excited about is being able to explain math concepts to you with diagrams. Language models are not quite there yet, but I think it’s going to—either with fine-tuning or some sort of specific training, I think we’ll be able to explain almost any math concept to you with a clever diagram. That’s much better than reading two paragraphs of explanation. So I’m very excited about that. I think we’ll be able to teach math significantly more effectively because of that.

Okay. Thanks a ton, Luis.

Thanks, Mark. And our last question comes from Wyatt Swanson at D. A. Davidson.

Hey, guys. Thanks for the question. I had a question on resurrected users that you called out. Could you talk a bit about how your initiatives have changed on driving users back to the platform? And could you maybe give some color as to how much engagement these resurrected users are driving?

Yeah. I mean, over time, we are shifting more and more work to people that we’ve seen before, because there’s just so many of them that lapse from Duolingo for six months and then come back. So we’re spending effort on that. And I think there’s a lot of room because we have just not worked on them as much. So I think there’s room on monetizing them better. There’s room on getting them to have better experiences coming back. I think that’s what I can say. We’re working on it and it’s getting better over time.

Okay. Fair enough. And then I had one on the intermediate English course. I know it’s really early, but just in terms of users enrolling, is it primarily new users that are coming onto the platform that find out that there’s advanced English, or is it users that are already enrolled in basic English? They see the intermediate now and they’re moving up.

It’s both. At the moment, I think it’s more of the current users, but that’s partly because we have done very little to tell people that there’s more advanced content. We’re going to be doing more over the next year and a half. So we’re getting both, but it’s more the current ones. Thank you.

Okay. We have no more questions. So I’ll just turn it back to Luis.

Well, thank you, Debbie. I’d just like to thank everyone for joining us and we look forward to seeing you at Duocon next month.