Q4 2024 Earnings Call Transcript

Good evening, everyone and welcome to Duolingo’s Fourth Quarter and Full Year 2024 Earnings Webcast. Today after market closed, 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 von Ahn, our Co-Founder and CEO; and Matt Skaruppa, our CFO. They will begin with some brief remarks before opening the call to questions. [Operator Instructions] Just a reminder that we will 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 in our filings with the SEC. The 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 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. Look, we had a wonderful 2024, and our results show it. But before I get to the details, I wanted to remind you all of the broader context. Since our IPO, which was 3.5 years ago, we’ve added 30 million daily active users and over 80 million monthly active users.

We’ve nearly tripled bookings, and we’ve gone from breakeven to a 25.7% adjusted EBITDA margin in 2024. All of this shows the power of our approach. We experiment relentlessly to improve user experience and monetization. Through gamification, social features and our social-first marketing, we scaled our reach, converted more free users into subscribers, improved learning outcomes and maintained strong profitability. And in 2024, we ended the year with an outstanding record quarter.

Daily active users hit 40 million, growing 51% year-over-year. We added more new subscribers than ever before and delivered our highest quarterly bookings, revenue and adjusted EBITDA. Our Q4 outperformance was largely driven by stronger-than-expected Duolingo Max subscriptions, including upgrades from current super subscribers and by continued momentum in our Family Plan, particularly during the first few days of our New Year’s promotion. I’m especially excited about the traction we’re seeing with Duolingo Max.

Since launching Video Call, our GenAI-powered conversation feature, user engagement has grown meaningfully. Max is now available to the majority of our DAUs and represents about 5% of total subscribers. We are still very early in driving Max monetization and believe there is a lot of room to grow. And like I said, our Family Plan is also delivering strong results.

It now makes up 23% of total subscribers and continues to show higher retention and LTV than individual plans. This year, we have three priorities. First, we’ll continue to drive subscription bookings by growing users, improving subscriber conversion and promoting Duolingo Max to more learners around the world. We’ll do this by running hundreds of experiments each quarter, and I feel very good about how quickly we are moving.

Second, we will leverage generative AI to improve the Video Call experience, which I’ll describe it more in a moment. We’ll also use GenAI to scale content across our language, math and music courses even faster. Finally, we’ll remain disciplined with our investments, balancing strong top line growth with measured progress toward our long-term profitability targets. I want to go into a little more detail on AI and scaling our content.

Duolingo Max is a key priority for us. Users are engaging more with Video Call, and we’ll work to make it an even better conversation experience. We believe Video Call will help us teach better, especially for our more advanced learners. Most of our AI costs are tied to Video Call, and they scale based on Max subscriber growth.

We’re prioritizing using the latest AI models to deliver a high-quality experience. What that means is that we’re not concentrating on cost optimization right now, even though we’re confident these costs will come down over time. AI and automation tools are also allowing us to expand content and courses faster than ever before. We believe this will help us reach more learners globally.

This includes also scaling our courses for Math and Music, which are showing strong early adoption. Today, these subjects have a combined 3 million DAUs, and we see a lot of room to grow, though as I’ve said before, this will take time. The market opportunity in both language learning and other subjects remains substantial, and we’re making strategic investments now in order to fuel growth for years to come. While this means a more moderate pace of profit growth compared to the exceptional levels of the past 2 years, we’re still delivering margin expansion, just with an eye on building something even bigger.

And with that, I will turn it over to Mr. Matt Skaruppa to walk through the financials of our outlook for 2025.

Thanks, Luis, and good evening, everyone. As Luis mentioned, we closed out 2024 with a record quarter, capping off another exceptional year for Duolingo. In Q4, we grew total bookings by 42% year-over-year. Revenue increased 39% year-over-year, and we expanded our full year adjusted EBITDA margin by about 8 points.

These results reflect the strength of our business model and our execution fueled by strong Duolingo Max adoption and demand for our Family Plan. We feel great about our momentum going into 2025. Now to our guidance. Our full year guidance has bookings growing 25% year-over-year at the midpoint or 27% on a constant currency basis. This growth will be driven primarily by subscription bookings, which we expect to grow around 31%.

Our guidance puts us on track to surpass $1 billion in bookings this year. For Q1, we expect bookings to grow approximately 28% year-over-year or 32% in constant currency, primarily due to continued strength in subscription bookings that are projected to grow about 35% year-over-year. This bookings guide is supported by continued strength in our DAU growth, which we expect to be in the mid-40s for Q1. Our guidance assumes prevailing foreign exchange rates. And as a reminder, over half of our bookings come from outside the U.S.

So every 2% increase or decrease in the value of the dollar versus our basket of currencies has about a $10 million headwind or tailwind, respectively, on full year total bookings. Our bookings guide includes Max bookings, and I want to remind you all of Max’ impact on gross margin. Max incurs marginal AI costs to drive its features like Video Call, and its higher pricing more than offsets these costs, driving increased lifetime value and gross profit, adjusted EBITDA and free cash flow dollars but a lower margin percentage compared to Super. In 2025, we expect a temporary 170 basis point year-over-year impact on gross margin, primarily due to Max.

In the first half of the year, there will be roughly a 300 basis point year-over-year impact as we prioritize rapid product innovation to drive Max adoption. We expect margins to improve in the second half of the year as we work to improve AI costs. As we scale, we remain committed to delivering both growth and profitability, and as we discussed, this year will reflect a more moderate pace of margin expansion following 2 years of exceptional gains. For 2025, we expect to expand our adjusted EBITDA margin by nearly 200 basis points to 27.5% as we continue to gain leverage across all categories of OpEx.

And we expect our incremental margin to be between 30% and 35% for the year. There are a few items we want to call out that are baked into the full year adjusted EBITDA margin outlook. The first is our investment in AI and automation that Luis and I had mentioned, which we expect to become more efficient throughout the year. Second, even though we plan to hire at about the same level as last year, we expect to hire faster and earlier in the year, allowing our new hires to ramp up and contribute sooner.

And finally, we’re capitalizing a bit less R&D spend than a year ago now that we’ve launched Max and several internal tools that enable us to create content significantly more efficiently. The impact of these investments on quarterly adjusted EBITDA margin will be the most pronounced in the first half of 2025. So for Q1, we are guiding to an adjusted EBITDA margin of 25%. In addition to the items I mentioned before, Q1 also includes a time shift of marketing expense, which will be more frontloaded than last year based on planned marketing campaigns. Now I’ll discuss how bookings and profitability will trend throughout the year.

We expect our Q2 bookings growth rate to step down from Q1 by about 3.5 points. Year-over-year bookings growth in Q3 should be about the same as in Q2 before stepping down in Q4, which will again be our biggest quarter in terms of dollar bookings. And we expect our revenue growth rate to step down throughout the year. As a reminder, we manage to an annual profitability target, and we may see quarterly variability based on timing of expenses, particularly AI costs.

But we are currently expecting Q2 adjusted EBITDA margin to be approximately 2 points lower than Q1 with meaningful margin expansion in Q3 and Q4 as we realize AI cost efficiencies and as our marketing spend as a percent of revenue comes down. We ended 2024 with a fully diluted share count of 49.5 million, and for 2025, we expect dilution of around 1%. And with that, I’ll turn it back over to Luis.

Thank you, Matt. Before we wrap up, I want to acknowledge something incredible. Our beloved mascot, Duo, has come back from faking his own death, thanks to the dedication of our learners around the world who completed 50 billion XP just in time to bring him back. And he is back to doing what he does best, pestering everyone to keep up their lessons.

And now we would be happy to take your questions. I’ll turn it back to Debbie to manage the queue.

Alright. Thanks, Luis. [Operator Instructions] Your first question comes from Bryan Smilek at JPMorgan.

Great. Thanks for taking my questions and congrats on the quarter. Just thinking about Max at 5% of paid subscribers, can you help us understand which cohorts are exhibiting the strongest growth by language or geo? And can you just help us understand how much of these are gross adds versus maybe shifting of subscribers from Super? Thank you.

Yes. So the – I’ll let Luis talk about some of the engagement we see because it is different between English learners and others in Max. But to your point around the 5% and where do they come from, the markets that they come from more or less look similar to Super markets, our Super Duolingo markets, with a couple of exceptions. Like for example, Japan has higher share of Max subscribers than Super subscribers, but it looks pretty consistent.

And then in terms of the cohorts, we saw really nice adoption for brand-new to subscription subscribers and also upgrades from existing Super subs. So we are excited that both of those played prominently in the growth of the subscribers up to 5%.

Yes. The other thing that I’ll add is that, as we mentioned in the last call, I believe, English learners in particular, really, really like this key feature for Max, which is Video Call with Lily. This is the only feature we have ever put in the app that gets used twice as much by English learners versus non-English learners. And that is actually translating in a slightly higher propensity to buy for English learners than Super.

Thank you.

Okay. Next question comes from Aaron Kessler at Seaport.

Thanks so much. Maybe just a follow-up to that question. Maybe on the AI investment, should we think about the AI investments this year as mainly Duo Max and Voice Calls or is there some other areas within AI? And then maybe just on the pricing strategy for English learners for Max.

Should we expect higher pricing as well for some of these maybe less developed regions? And then just what is your – how are you thinking about – or is maybe just – you expecting higher paid conversion rates versus trying to sell on Max there? Thank you.

Okay. So there is a bunch of things to say here. First of all, AI. The AI investments were – they’re basically coming in – and expenses.

They’re basically coming in two parts. We’re definitely investing in AI to automate things inside the company. What that does is that actually decreases costs. That’s good.

And it also allows us to go way faster, particularly in content creation. So the amount of content we’re able to generate compared to 2 years ago is way more. I mean, it’s like 10x or more than that. So that’s one thing.

The other big part is in just feature experimentation or kind of live features like, for example, the Video Call feature. That adds expenses because we have to query the large language model kind of in real time. The way we’re seeing it is, for now, we’re not even trying to optimize those costs. We’re just trying to go as fast as possible in generating the best possible features.

We know that these costs can be optimized in a lot of ways. And even if we do nothing, the cost will be optimized because the cost of querying large language models is coming down kind of every month. So the way to think about it is there’s going to be kind of an upfront cost for AI, which will happen probably throughout this whole year, but it’s because we’re in a unique opportunity, a unique time in history where we really want to develop the best possible AI features over time. So that’s with AI.

Now you asked about pricing, particularly for kind of probably the AI packages. At the moment, we have really put Duolingo Max in most – every country. There is a couple of countries where it’s not at. For example, it’s not in China simply because we cannot serve things from OpenAI in China.

So we’re going to have it there, but it’s not there yet. But it’s basically in every country. In some of the poorest countries, our price is too high is the truth, but because what we’re doing is we’re pricing it so that we never lose money on Max, and so that’s good. And that – for most countries, that makes sense and it is a price that makes sense.

But for example, in India, the price, I believe, is $70 a year and that’s too expensive for India and we know that. But we expect that, over time, we’re going to be able to bring that down, always with a caveat that we’re not going to lose money with it.

Thank you.

Our next question comes from Ralph Schackart at William Blair.

Good evening. Thanks for taking my questions. Just can you give us a sense on DAU growth, just a sense of how broad based that was? Maybe how is international doing relative to the overall reported rate?

And then just on gross margin, it sounds like this is sort of temporary in nature just to take advantage of that market opportunity. But Matt, longer term, is there anything structural here that you think you won’t be able to get back to historic margins? Any sense on that would be great? Thank you.

Yes. Let me talk about daily active user growth. So we’re very happy with the daily active user growth. I mean, we ended the quarter with 51% daily active user growth year-over-year.

It really – every region is growing. Now of course, it’s not true that every region is growing equally, but every region is growing. One thing that is really interesting to say is that the growth rate per region is not really correlated with how mature each region is. So for example, our – probably our most mature region is Latin America, but it’s also growing very fast.

That’s actually growing at like 80% year-over-year. So we’re – what that tells us is that we really are far from saturating any of our markets. We’re just – we’re growing pretty fast in all of our markets. I’ll let Matt talk about the profitability or margin expansion.

So Ralph, as I mentioned in my prepared remarks, it is temporary, as you said. It will be more noticeable in the first half of the year, as Luis mentioned, because we’re just going as fast as we can and we’ll optimize it in the back half. Because of that, by the back half of the year, I expect gross margins to return back so that they are roughly comparable where they’ve been in the past. But I don’t want to understate the fact that there is a structural difference, right.

There is a marginal AI cost. It’s just that our belief, and we’ve seen this in the market already, is that, that cost will come down such that it won’t be that different. So I feel very good about where our gross margins will end up as we optimize those costs.

And we are still going to have market – margin expansion, it’s just lower – at a lower rate than in previous years.

Alright. Thank you.

Thanks, Ralph. And next question comes from Ryan MacDonald at Needham.

Thanks for taking my questions. Congrats on a great quarter. Luis, it was interesting to hear the updated disclosure on Music and Math in terms of 3 million daily active users. I’m curious, given the scale now, how should we think about, one, maybe use of AI to further enhance the content in those two categories?

And then second, how we think about, Matt, for monetization over time beyond sort of maybe a Family Plan or another area? Thanks.

Thank you for the question. Okay. So we’re very excited about Math and Music. Like we said, we now have about 3 million daily active users studying Math and Music.

For your question for AI – and by the way, I should also say, those two are – those courses are growing faster than our language learning courses. So we do expect that these will continue being a larger and larger fraction of our whole pie. Now in terms of uses for AI, for Math in particular, AI is really going to be amazing. What happens with Math is, at the moment, we don’t have a lot of content in our Math course.

We have the equivalent of probably third to the fifth grade roughly of content. But of course, there’s a lot more we want to put. We want to have all of K-12 and maybe even some college math in there. Now the generation of math content historically has been pretty slow because you kind of have to make new exercises for every topic.

So for example, we make all these exercises that look really amazing for teaching you the coordinate system, but those kind of don’t help all that much for teaching you probability because they just look completely different, and they don’t help all that much for teaching you logic, etcetera. So you kind of have to make very different types of exercises. Somebody here said, math is like a combination of 1,000 subjects. It’s just a bunch of different things.

And historically, the first version of the large language models, were not very good at math. But of late, we really – the large language models have added reasoning. So they’re actually pretty good at math now, and I think that’s going to really accelerate how much content we put in there. So you’ll see over the next year or so that the pace at which we are adding content to the Math course is going to improve quite a bit.

And that, of course, will help us have a lot more uses. For Music, we’re also going to be adding a lot of content, but AI is probably not as transformative as it is for Math. And now in terms of monetization for these, at the moment, they are being monetized. I mean, they’re being monetized in the same way that we monetize our language courses.

Basically, when you do a math lesson, at the end, you see an ad. And if you don’t want to see ads, you can pay to subscribe. Also, you have – every time you make a mistake, you lose a life and if you want to get rid of that, you also can buy Super for that. So they are being monetized in the same way.

I think for the time being, we expect the monetization to remain the same. So one way to think about how much money Math and Music are making is roughly the proportion of daily active users. It’s not precise, but that is kind of one way of thinking about that.

Excellent. Thanks a lot.

Thanks, Ryan.

And our next question comes from Ross Sandler at Barclays.

Great. Thanks, guys. Just two quick ones. Going back to the Max penetration, the 5%, could you talk about like are all the English corridors like about the same in terms of their penetration or is there like a high watermark within there?

And are there non-English corridors that are working yet for Lily and the Video Calls? And then the second question is, the new chart you put out on the course units published, that’s pretty interesting. Kind of shows the ramp of content you’re talking about. How should we think about just like longer term how that correlates with other KPIs in your business like engagement or subs or that’s a cool chart, but how should investors think about that translating back to the business metrics?

Thank you.

Great. So in terms of Max penetration, yes, it’s 5%. It is, of course, not even in every country. Like Matt said, it is pretty correlated.

It’s not precise, but it’s pretty correlated with Super penetration. So generally, you see that wealthier countries are more penetrated than less wealthy countries. There is one difference, which is English learners are a little boosted up compared to Super. So for example, Japan is a prime country that is both wealthy and has English learners, so that’s pretty highly penetrated.

I mean – and by pretty highly penetrated, we’re still in the single digits there, but it is higher than the 5%. So that’s what we’re seeing. But it’s not just English learners. I mean, we really are seeing kind of learners of every language.

It’s just it turns out that English learners are a little higher than what you would see for Super. In terms of course units published, we’re super excited about the fact that we can now just publish content way faster, and it is because of AI. We’ve automated our content pipeline, and we are just going way faster. How that translates to KPIs is basically we’re going to have – not only are we adding more courses.

So what that graph shows is basically the amount of content we’ve been able to publish in each of the previous years. Some of that content goes to existing courses. That’s usually to more advanced sections. And some of that content is to brand-new courses.

Brand-new courses increase our usage. So for example, we may not have had – for Korean speakers, we may not have had a Spanish course. But now we have a Spanish course for Korean speakers. That’s just an example.

So now anybody who wants to learn Spanish from Korea, it used to be the case that they needed to learn through English. So they needed to kind of already know English in order to learn Spanish. Now they can learn it just from Korean. So that should increase our total users.

And then the more advanced sections should get – also should increase our total users because it should get the more advanced learners there. A large chunk of the content that we’ve been publishing is in more advanced – for more advanced English learners. And that’s something that we’ve talked about quite a bit. But we’re very excited about the opportunity for more advanced English learners.

Okay.

Thanks, Rodd. Next question comes from Edison Kai [ph] at CITIC Bank.

Hello. Congrats on such an excellent quarter. And I have question related to the – for the past quarters, you’ve mentioned about how the growth of new users and resurrected users growth in the past few quarters. So what do you think of what we’re doing right this quarter about like the marketing campaign events you are having?

So are they calling for more new users or are we calling back those resurrected users? And for the markets you mentioned like Japan and Latin America, we see the growth from, I guess, new users and resurrected users. So what do you think of like the boost of these two – like two user growth regions and which one do you think will be more like effective way to grow in the next few quarters? Thank you.

Yes. This is a really good question that’s important in order to understand how our business works. Typically, when people stop using Duolingo, they – it is rare that they stop and never come back. Most of the people who stop using Duolingo stop for a few months or for a year, and then they come back.

So we have a lot of people who are resurrecting, and resurrecting, we mean they haven’t been around for at least 30 days. Of course, as we get more mature, the fraction of kind of top of the funnel that is resurrected gets higher and higher because, for example, when we had just launched Duolingo, there was no users to resurrect. So, all users were new. And as we get more and more mature, a higher fraction of people are people that are coming back.

And that is pretty consistent with how mature we are in each country. So in a country that we’ve been operating for a lot longer, like the U.S., you would see a higher fraction of resurrected users when compared to new users versus in a country where we’ve been operating much less time, for example again, India, where we really didn’t start spending effort in India up until a couple of years ago. So there, we’re still seeing way more new users than we are seeing resurrected users. At the moment, if you look overall, we are seeing slightly more resurrected users than new users at the top of the funnel.

But those numbers are pretty similar. It’s just slightly more for resurrected users. And our marketing in general, most of our marketing goes into kind of these social media campaigns like what you’ve seen kind of on TikTok or YouTube or Instagram, etcetera. And those are actually really good at both bringing new users and resurrecting users, and we know that because whenever somebody resurrects or whenever a new user comes in, we ask them kind of where they came from.

And we know that our campaigns are good at doing both. So we’re doing that. Now, in terms of work for them, the one piece of work that we do still need to work on is we know that the retention of resurrected users is not as good as the retention for new users, and we need to improve that. So, we are working on that.

Part of the thing that happens is when somebody has been gone, a lot of times, they have been gone for a very long time. We should – what we should do is we should assume that they have forgotten everything. If they have been gone for 2 years, we should assume they have forgotten everything. We are not.

And so that’s something that we could just do better.

Okay. Thanks Luis.

Thank you.

Okay. Next up is Andrew Boone at JMP.

Thanks for taking the question. I want to go back to Ross’ question. It’s a derivation of it. Luis, in your first priority, you talked about growing users, improving subscriber conversion, and then promoting Max through testing, right?

And so like the question is, okay, you guys are doing so much more content. How does that relate to testing in terms of generative AI? Are you guys seeing accelerated like content cadence in terms of what you guys put out there? Help us understand your testing velocity as it relates to 2025?

Thanks.

I think I understand your question. I am not 100% sure. But basically, if you are asking about just the types of tests that we run, I mean we usually run these A/B tests to try to improve our core metrics. And we have to concentrate on a number of core metrics.

In this case, we are concentrating on getting more users to subscribe to Max. The types of tests that we run are things like when do we advertise Max, who do we advertise it to, do we advertise to subscribers, how often do we advertise it to subscribers, what do we say when we advertise to them. And we have found some vectors that really allow us to experiment a lot. So, for example, one of the main ways we are getting people to subscribe to Max is with an ad for Video Call.

We show kind of what it looks like to do Video Call, and that is very effective at getting people to subscribe. And in terms of the velocity of these tests, I feel really good. We are – well, in 2024, we ran more tests than we have ever done before. And I believe that in 2025, we are going to run way more tests than in 2024. So, our velocity is looking really good and we have a really good roadmap for that.

Yes. Hopefully, that answers your question.

Thank you.

Great. Next question comes from Justin Patterson at KeyBanc.

Alright. Thank you. It’s fitting I am going to ask a bootleg Jack Antonoff a question wearing a bootleg Duolingo T-shirt. But maybe building on just Ross’ and Andrew’s question a little bit more.

Luis, you have a mission to make education more affordable to the masses. You alluded to it earlier with Max. You don’t want to be pricing at bad unit economics internationally. So, as we get into this environment where inference costs continue dropping, how do you think about just the pace that you might provide some of those savings back to users?

And then secondarily, just thinking about Video Calls, any additional learnings you can share about just the proficiency of people engaging there since you have already done the hard part. You have gotten casual users to jump into the category. I am curious how well you are doing converting them from, say, more casual usage to advanced proficiency. Thank you.

Yes. The – sorry, what was your first question again? Just around proficiency?

Yes. First question was just as inference costs continue to drop, how do you think about the value of…?

Yes, generally, we have these three tiers. We have the free tier. We have the Super, and we have the highest tier, Max. At the moment, because of inference costs, particularly for Video Call, we have to have Video Call in the highest tier.

That’s not necessarily going to always be the case. It may be the case that, at some point, it gets cheap enough for us to put in a lower tier. We may even be able to give some of it for free. And in general, we will be testing whatever is best for the long-term, not just LTV, but the long-term health of our app.

And we know that, in many cases, giving things for free is actually the best thing for our app. So, we are just going to be testing a lot. At the moment, our hands are a bit tied because the costs are just high – too high, but we will definitely be testing where to have these features. Now, in terms of getting the users proficiency, we are really happy with how much better we are teaching, particularly conversation now with Video Call.

We know that before this, we really didn’t have a great way to practice conversation and now we do. And that is – it’s working. I mean it’s really effective. Thank you for calling me bootleg Antonoff.

Thank you. I am going to save Mr. Skaruppa for another day.

That’s going to be a thing.

Alright. Next question comes from Mark Mahaney at Evercore.

Okay. Thank you. You talked about, in the letter, making Lily more dynamic and interactive. That seems like a pretty low bar to me.

So, what are you thinking about? Secondly, Matt, the subscription revenue per average sub, so we have sort of – it seems like we have hit this inflection point. I assume it’s relatively – like in that guidance that you gave for the full year, should we assume that, that subscription revenue per average subs continues to grow because of greater Max and Family adoption? And then third, the ad revenue, I guess was a little – came in a little lower and now you have pressured gross margins a little bit.

Is that just because it’s less of a focus for the company? You just don’t need the ad dollars, or I don’t know, is that a problem you need to solve? Thanks a lot.

Thanks Mark. Yes, Luis, you can take the first one. I will take the other one.

Yes. I mean in general, what we meant by that is, at the moment Lily is not necessarily your best friend yet. We want that to happen. We really wanted to have – we want it to be the case that you want to talk to Lily.

And by the way, that is the thing that really differentiates us from every other learning thing. Not only do we teach effectively, but also people actually want to use Duolingo. And in particular, for Lily, we want people to actually want to talk to her. And so that just means every time you go there, she will have something interesting to tell you.

She will remember what you told her last time. She will ask you about your problems, etcetera. And we are working on that. I mean it’s a fine balance because one of the things that is hard is we are getting these people who may not be able to express themselves all that well.

Yet we want to be able to ask them how their lives are and tell them something about Lily’s life. So, that’s what we are going to be working on. And there is a lot of work to be done there. But I think by the end of the year, what’s going to happen is there is just kind of – we are just going to see much higher engagement numbers.

Even though they are already – I feel pretty good about our current engagement numbers with Lily, we are going to see much better engagement numbers with her.

And then on the ARPU question and on the ads, on the ARPU, you are absolutely right, Mark. We told you last call that we thought it would be flat to up, and it ended up, up, which was great. And it gives me a chance just to remind everyone that we talk a lot about pricing, and we run pricing experiments all the time. We are running them now, price point experiments.

But the larger shift we expect over the next several years is we are effective at continuing to shift people to Family plan and to shift more folks to Max. That will be the larger ARPU swing. For the rest of the year, you are right. The guide assumes that we see positive ARPU year-over-year throughout the year.

And then on ads, you are absolutely right. We are a subscription business, and so our – most of our energy is running experiments to optimize our subscriptions. Ads was lighter in Q4. It’s a combination of volume and RPMs. The volume is a good new story for us because we are showing more Duolingo ads to get folks to take Super or Max.

So, I don’t expect ads to fundamentally shift up a gear. We are focused on subscriptions.

Thanks Matt. Thanks Luis.

And our next question comes from Wyatt at D.A. Davidson.

Yes. Thanks guys. I just had a quick one here. Just given the massive TAM of 2 billion language learners, it’s still really early, clearly, of the potential user base that you guys get to have.

Can you maybe talk about like where you see user growth coming from next? Like what markets or demos do you consider the next growth avenue? Like what are you under – where are you underpenetrated right now? And like how are you setting yourselves up to achieve growth in those areas?

Yes. Thanks for the question. Okay. So, at the moment, we are seeing really growth in every geography.

They are not identical growth in every geography, but we are seeing growth in every geography. There – one of the things I said in a previous answer, the – what’s amazing, at least to me, is that the – how penetrated a geography is or how mature a geography is, is essentially not correlated with how fast it’s growing. So, some of our most mature geographies like Latin America are some of – also some of our fastest ones, which means we are just – there is really – we are nowhere near tapped out. So, that’s good.

Now, in terms of what are – where are we more or less penetrated, generally, Asia is the place where we are least penetrated, simply because we started much later there. And so I think there is a huge opportunity there. Places like Japan, Korea, India, China, I think there is a huge opportunity there. What we are doing there is we are setting up marketing.

I mean we have already set it up, but it’s a little less developed than our marketing in some of our more mature markets. But we are setting up marketing. And by marketing, I mean just to be clear, don’t mean a bunch of performance market or anything, essentially very similar marketing to what you see here in the U.S., which is a lot of social media stunts and stuff like that, that work really well. Fun fact, by the way, Duo, our owl, faked his death in every single market that we had except for Japan because it turns out that in Japan, joking about death is not as kosher.

So, in Japan, he was just not dead.

Got it. Appreciate it.

Alright. Moving on to – we have got Chris Kuntarich from UBS.

Great. Thanks for taking the question. I just wanted to ask you about the four days of stronger promo period and maybe just zoom out for a second and kind of, you hear about how did Max perform versus your expectations in the promo period? Are you seeing a larger uptake for Max versus the Super plan? And just kind of how does that mix look as we look at adoption during that promo period? Thanks.

Yes, I am happy to take a shot and Luis can come in. So, the overall promo period surpassed our forecast. And it did so both on Max, as you kind of alluded to. Max did better and then the Family plan also did better in those last four days of the year.

So, it was kind of a well-rounded outperformance in those last four days. I do think that it’s important to know that the performance throughout the quarter was strong, though. It wasn’t just the last four days or the promo period. We had strong user growth throughout the quarter and a bunch of other bread-and-butter monetization experiments that really paid off in the quarter, a bit even differentially and above our expectations.

So, it wasn’t just the New Year’s promo.

Thanks guys.

Alright. Thanks Chris. Next question comes from Georgia Anderson at Wolfe, stepping in for Shweta.

Hi. Thanks for the time. I was wondering if you could talk a little bit about the languages that are available on the app, on the Duolingo Max tier. I think most of the Romans languages were available as of a couple of months ago, but I was wondering if there is kind of an ideal number of languages, what the rollout of that has kind of been, languages are kind of easier or more difficult to add.

Just anything along those lines would be helpful.

Sure. Happy to take that. Generally, we have our largest languages by demand already available in Max. So for example, English available, French, Spanish, etcetera.

Japanese is available. Chinese is available. Now, the thing to understand about our languages is that there is a pretty lopsided demand. So, for example, about half our users are learning English.

A little under 20% are learning Spanish, and a little under 10% are learning French. So, those three combined is like 75%, give or take. And that’s just three languages. And then all other languages combined, we – in total, teach about 40-some languages.

All other 37, call it, languages, only account for 25%. So, we cover – with Max, we cover at least the big bulk. We have – we cover the top eight. It is true that it is harder to add it to non-Romans languages.

So, for example, it’s harder to add it to Japanese or Chinese. And part of what’s difficult there is that it’s just learning those languages, particularly for people from Western countries, learning those languages is so much harder that when we put, for example, a Video Call on – for an early Japanese learner, they really – it’s very hard for them to understand anything. Whereas if we put a Video Call in Spanish, if you are in the U.S. and we put a Video Call in Spanish, a vast majority of people, even if we haven’t taught them anything, kind of understand like [Foreign Language]. Like most people can do that, whereas for a language like Japanese, it is much harder in that respect, so we have to do all kinds of things to make it – to really either put things like Video Call, start them later or dumb it down even more than we would need to.

That’s the type of stuff we have to do.

That’s helpful. Thank you.

Alright. Next up is Curtis Nagle at BofA.

Awesome. Great. Thanks so much for taking the question. Maybe, Matt, one for you, and I guess forgive me if I misread this.

Just maybe expand on why the revenue growth is expected to step down a little bit in terms of rate of growth, right? Max is ramping. And then as a related question, where do you think Max, maybe for you, Luis. Where do you think Max penetration could end up by year-end?

Kind of what’s embedded in the guidance? What are your aspirations for that?

Thanks Curtis. Yes, on the revenue guide, that’s just a factor of the math of the bookings rate. So, revenue in any given quarter is essentially just a composite of the last four quarters of bookings growth rate. So, that’s just a mathematical outcome of the bookings guidance.

So, I don’t think there is all that much there. I would just also call out that as we lay out the bookings guide for the year, that does have a pretty sizable FX impact, right. So, that will be felt over time through the revenue guide as well, so both of those things are at play. And then I think Luis had the second part of the question.

Just how high Max penetration can get, I mean look, in general, we don’t really know where that is – we know it can get higher than 5%, because we are seeing it grow kind of on a daily basis. So, it will get – it will grow throughout the year, but we don’t really know how high it is. I am sure we have something in the model.

Yes. I mean I think our expectation for Max is that it’s a material amount of bookings in 2025, because we are seeing really nice growth in penetration in it. But as Luis pointed out, we think it’s – we are early days. There is a ton of rapid innovation to do on those features and how we surface that and how we price that and then what the AI costs.

So, it’s just very early, and we are excited about it for that reason.

Okay. Thank you.

Next question comes from John Colantuoni at Jefferies.

Thanks for taking the question. I wanted to ask one about reinvestment. When comparing how users typically interact with Duolingo today relative to intermediate English learners, how do you expect a growing focus on intermediate English learners to evolve your approach to marketing and flexibility to reinvest dollars back into product and innovation? Thanks.

Yes. In terms of the more intermediate, so just to remind people, I think take a little step back. We have decided to concentrate on more intermediate English learners as of relatively recently. It’s really a couple of years ago.

The reason for that is because that is a huge part of the market that we pretty much were not serving. We had English courses, but they were mostly for beginners. So, we started really investing in more advanced English courses. We now have the content there.

So, the content is now there, and so it’s kind of a matter of time for people to start interacting with this content. And one of the things about Duolingo is that our main way of growing is word of mouth. That’s how we have mainly grown. And for the last more than a decade, basically, the word has gotten around that Duolingo is a fun way to learn a language and people have come to the app.

In English learning markets, so in non-English speaking countries, the word also got around that we are good for beginners, but not for advanced learners, because the content wasn’t even there. So, at the moment, if you go and talk to your average person in – pick any country where they are trying to learn English, many of them will tell you, yes, if you are a beginner, that’s what, you should use Duolingo. If you are more advanced, that’s not the case, because they don’t know that the content is there. It is, of course our job to make sure that they have the content there.

Because most of what we do is word of mouth, we expect that through word of mouth, it will get out that we have the content. Now, the problem with that is that it will take some time for that. We will do some marketing for that, and so that will help. But one thing that is just maybe a challenge is that our most effective marketing throughout the world is things like Duo deciding to fake his death.

That’s very effective, but that does nothing for convincing people about whether we are good at more advanced English or not. It’s just good to get the word out. So, we do have to change some of our marketing tactics for that, and we are working on it. But – so you will basically see us do that.

My – what we are seeing is that the content for more advanced English learners is getting more and more interaction. So, we are getting more and more advanced English learners over time. And so that number is growing quite a bit, and my sense is that by the end of the year, we will have way more. And then by the end of the year – the year after that, we will have even more.

So, it’s just growing in pretty nice pace.

And John, just to – we are spending – it’s not just on English learners, but we are spending more in absolute dollars in marketing this year than we did last year. And we are still getting leverage on the S&M line, right, because those growth rates are just smaller than revenue. So, I don’t – our reinvestment rate in marketing, I feel very comfortable with. We are increasing marketing spend and yet still getting leverage.

And I think that’s a nice spot to be – pretty lucky spot to be in.

Thanks. Appreciate it.

And the next question comes from Alex Sklar at Raymond James.

Luis or Matt, just you commented on higher Max LTVs that you have observed so far. Obviously, the 2x pricing is a big piece of it. But I am just curious, how has early Max retention relative to Super? And then given some of the higher usage features like Video Call that you called out, how are you thinking about retention of Max over the medium-term?

Yes. The retention so far looks good. But as your point – you pointed out, it’s early days. So, we don’t actually have the full on, but we have ways we estimate that and we feel good about it. And the LTV is really strong. So, Max is undeniably our highest LTV tier, and the more we can shift there, the better. Hopefully, retention goes up over time as well as we improve the features.

Alright. Great. Thank you.

Okay. We have got Eric Sheridan at Goldman Sachs.

Thanks so much for taking the question. Maybe, Matt, coming back to the way you framed incremental margins for 2025, wanted to maybe ask a two-parter. First, what do you, as a team, see as sort of the critical investments that have to be made that would put somewhat a pressure on incremental margins? I think you called out hiring in the prepared remarks, but I want to put a little bit better scope around that.

And in prior years, you have exceeded margin guidance, and it’s proved to be more conservative or there has been outperformance. What could be areas in the business where you could outperform the margin guidance as opposed to it more coming in line with the expectations as you go deeper into the year? Thanks so much.

Yes. Thanks Eric. No, it’s a great question. So, just to reiterate what I said in the prepared remarks a bit about the less important spend items, the one we are spending the more money on earlier is sales and marketing, right.

And we are hiring faster. So, that’s what you mentioned. We want to do those things because we see an opportunity, and those are helpful. But the biggest, most important investment and increased cost that’s leading to our margin expansion of 200 basis points is the marginal cost of Max.

And so that’s a really important driver of margins, but it just means that Max is going to grow nicely. And that’s great for the top line. And it’s great for the ultimate LTV of the platform because, again, as we have said on this call and we said in the shareholder letter, we believe over time that those costs will come down for Max. But now is not the time to focus the teams on cost optimization.

It’s the focus that – now is the time to focus the team on going as fast as possible to grow adoption of that tier. So, that’s really the biggest piece of the puzzle on the margin side, Eric. And the upsides to it are costs could come down faster on AI costs. But again, I still feel really good about where our incremental margins are.

They are in our long-term margin range. And so I feel very comfortable with how we are balancing margins as we are growing really rapidly on the top line.

Great. Thank you.

Okay. And it looks like final question is from Nat Schindler at Scotiabank. Nat, you are on mute.

We can’t hear you, Matt. Maybe your headphones because you are now on mute and you were not before. No.

Alright. Sorry about that, got to love technology. Right, I am showing no further questions, so we can go ahead and wrap it up. I will turn it back to Luis.

Thanks Debbie. I would just like to thank everyone for joining us and we look forward to seeing you on the next call.