Duolingo, Inc. (DUOL) Earnings Call Transcript & Summary

Good afternoon, everyone, and thank you for joining Barclays' TMT Conference. My name is Mario Lu and I'm on the Barclays research Internet team. With us today is Luis von Ahn, CEO and Co-Founder of Duolingo. Welcome to our conference, Luis. Very excited to have you here.

Hi, Mario. I'm very excited to be here.

Great. So yes, this is Duolingo's first time at our conference, so I'm sure many people are familiar with the app itself. But can you give us a rundown of the business? What the current revenue breakdown is?

Sure. So Duolingo, what most people know us for are our main product, which is the most popular way to learn languages in the world. It's an app that you can use on iPhone devices or Android devices. To give you an idea of the scale, there are more people learning languages on Duolingo in the United States than there are people learning languages across all U.S. high schools combined. We teach a lot of languages. We teach all the big ones, of course, that you've heard of, like Spanish and English and German, et cetera. We also teach some smaller languages like Irish, or Klingon from Star Trek, or High Valyrian from Game of Thrones. And we've grown a lot. We currently have about 42 million monthly active users, and this growth has been almost entirely organic. So basically every year, we've just grown from the previous year, and it's all been organic basically through word of mouth. And there's 3 reasons why I think we've grown so much organically. The first one is that we were mobile first. So when we launched Duolingo, this was about 10 years ago, and at the time, most companies were spending most of their effort on their website. Everybody knew mobile was going to be big, but the biggest thing was still the websites, whereas we decided to spend most of our effort on our mobile apps. So we launched very high-quality apps. And because of that, for example, Apple chose us as the iPhone App of the Year when we launched, et cetera. Another reason why we've grown a lot is because we're free. So it's entirely free to learn on Duolingo. You can learn as much as you want on Duolingo entirely for free. And because of that, people download it and they use it. And they don't have to pay, and then they tell their friends. And another reason we've grown so much is because it's a lot of fun. So we realized early on that the hardest thing to learn a language -- the hardest thing about learning a language is staying motivated. So we spent a lot of effort making Duolingo feel as much like a game as possible. So this is why people, when they download it, they spend a lot of time on it. And again because of that, they kind of tell their friends, so that's the idea. Now in terms of how we make money, we -- for the first several years of Duolingo, we actually were not monetizing. So we launched, and this was just a free way to learn languages, and we were just working on making our product more and more engaging to grow. About -- in about 2017, so give or take 4 years ago, we decided it was time to start monetizing. And the way we ended up deciding to monetize is not the most straightforward way that educational things monetize. So usually if you want to monetize something in education, the more straightforward is just to charge people to learn. We don't do that, and in particular because we're a very mission-driven company and we really want to give equal access to education to everybody. Secondly, we have been a free way to our language for several years. We thought it would be silly to just start charging for people to turn a language at that time. So we decided to monetize in this freemium model, which is kind of similar to what Spotify does or what some of the dating apps do, which is that you can use Duolingo as much as you want entirely for free. But if you don't pay us, you have to see some an ad at the end of the lesson, and we make money from the ads. And then also, if you don't want to see the ads, you can also pay us to subscribe to the premium version, which is ad free. And also it has a few extra perks. So that's the idea of how we monetize. Today, most of our users use us for free. So about 5.5% of our users, of our monthly active users, are paying subscribers. And that number keeps going up. So a year ago, that was 4%. And so it seems to be going up by about 1 to 1.5 percentage points per year, the fraction of users that are paying subscribers. So that's the idea. And you asked in terms of kind of our split of revenue. Today, we make about, call it, 70% of our revenue, 70%, 75% of our revenue comes from the subscription. About 10% of our revenue comes from ads. And then about another 10% comes from a completely separate product called the Duolingo English Test, which we can talk about more later, but it's a completely separate product. That's the split of our revenue.

Awesome. That's a great overview. And then for those that are a little bit less familiar with the story, can you kind of walk us through how you define your TAM and Duolingo's current share in that market? And then how do you see your addressable market change over time as you enter new categories, such as literacy and math?

Yes. Okay. So I mean the first thing to say, because you asked about other subjects, when we started Duolingo, what I wanted to work on was something related to education. That was the idea. And in my head, just education is very general, so what we decided to do was to start by teaching one thing. But the goal was never just to teach one thing. It was to start with one thing and then teach other things. It turns out for the last 10 years, we've been teaching one thing, which is languages. We decided to start teaching languages because it has a very large TAM. And it's not just TAM, it's just a large number of people are learning a language. There's about 1.8 billion people in the world learning a foreign language. And combined, they spend about $60 billion a year, so that's kind of how much is being spent on language learning. Now there's a lot of things to say about that. The first is that the market, this market, this $60 billion a year, is to this day still overwhelmingly off-line. So the way I think of this market kind of in my head, the prototypical user or a person that's spending in this market is somebody who's in Brazil taking night classes to learn English. So that market is mostly off-line, but it is switching online, and we see that the online part is growing just a lot faster. And that our hope is that as the leader in online language learning, we're going to be able to capture a lot of that shift. So that's kind of what we see. Now In terms of teaching other things, we've been mainly teaching languages for the last 10 years, but a lot of the technology that we've developed can be used to teach other stuff. So we're starting to expand beyond languages. So for the last maybe a year or so, we launched -- about a year ago, we launched a literacy app. So it's called Duolingo ABC. That's an app to teach young kids how to read in their native languages. It's not foreign languages. It's just you're already -- you're a kid. You already know how to speak, but you don't know how to read. You can use our app to learn how to read. That's called Duolingo ABC. It's growing a lot. We're working on an app to teach elementary school math. That's -- we're going to launch that next year. Now for the next couple of years, we're overwhelmingly going to be -- our business in terms of our revenue is going to be mostly about language learning. But I'm hoping that in 5 years or so, language learning is just one of the verticals that we have, and we have -- our other businesses have grown a lot. Now you're asking our numbers. Our revenue numbers are in our latest earnings call. But this year, we're aiming, at least for bookings, we're aiming at high $200 million plus.

Got it. Okay. And then you just touched a little upon this but you mentioned potentially the math app may be rolled out sometime next year, monetization probably further down the road. But in terms of like the algorithms or things you developed already on the language side of the business, like how transferable is everything to other subjects, such as math? And then in theory even those later down the line, should we expect these ramp into other categories to be a little bit faster because of these transferred learnings?

Yes. So okay. So I'll say -- so in terms of technology, a lot can be transferred. So we have a lot of pretty sophisticated technology that a lot of people sometimes don't realize -- I mean I'm sure a lot of people that are listening have used Duolingo before and they're users of Duolingo. When you see Duolingo, it's very cutesy and it kind of looks -- it looks like a game, and it's so gamified. A lot of people don't realize there is quite a bit of sophistication in the background. So for example, whenever you start a lesson on Duolingo, that lesson is entirely personalized to you. So we actually -- whenever you're using Duolingo, we watch you -- we watch every exercise that you get. We know whether you got that exercise right or wrong. If you got it wrong, we know why you got it wrong. And then we use all of that data to build a model for each user. So for you in particular, we may know you're not very good at the past tense, for example. So whenever you start a lesson, we will give you more exercises that are related to the past tense. And it's certainly a little more sophisticated than that. If we were only to give you exercises that are related to the things you're not very good at, those lessons would be just like lessons from hell. They would be very difficult. So we actually -- whenever we start a lesson for you, we're trying to optimize for 2 things: We're trying to teach you better, so we would give you more things related to the past tense if you're not good at the past tense. But also, we're trying to give you things that we think you have a chance of getting correct. So we, for every exercise that we're about to give you, we know the probability that you have to get that exercise right. And so we try to give you things that you have about an 80% chance of getting right. And all of that sophistication is something that can be transferred pretty easily from languages to other things that can be learned through a lot of drills, like math or literacy and stuff like that. So we have a lot of technology and a lot of the gamification that can be transferred easily. So I strongly believe we can create very good products to teach these other things. That, to me, is not -- I'm like -- I'm pretty sure of that. Now when you asked about speed, the one thing that I think people need to understand, that I want investors to understand about us, is Duolingo grew organically, and we love that. We love that it grew organically. That really shows you that the product is very, very good because the way we grew is by having people tell their friends. That's good. There's another way of growing products, which is basically through marketing. There's nothing wrong with that, but that is just not the way we do things. We prefer to grow them through the product being so compelling that people tell their friends. If you grow things organically, the one downside is you don't grow from 0 to 50 million users in 2 days. You could do that if you were to launch a massive marketing campaign, and so you could do that. But when you grow organically, your rate of growth is just kind of how fast people tell each other. And so there's that. So what you're going to see for our monetization for our other products is we're going to do something similar to Duolingo. Now of course, I think it's going to be faster than Duolingo. It took us from where we -- when we started to making, what, call it, $300 million a year or whatever, it took us this -- the last 10 years. I think we can do that much faster, but I don't think we'll be able to do that in 6 months. Just because, first, we're going to get it to grow organically. And then once we have enough of a user base, call it, I don't know, 1 million daily active users or something like, that's when we would turn on monetization. So in terms of when these other products are going to be meaningful, at least in terms of revenues for our business, you should be looking at a 3- to 5-year time frame as opposed to a 1- to 2-year time frame.

Got it. That's helpful. And then to that point, I think it's pretty well understood, like you said, that Duolingo is a product-first company rather than a marketing company. You guys have roughly 70% of employees focused on working on the app itself. With that being said, you had success of late on the social media channels. So do you think this is an opportunity for Duolingo going forward? And I think you kind of already answered the second part of the question. And do you think marketing spend is necessary to drive user growth over time?

Yes. So this is a good question. So this is our marketing initiatives. I mean -- so yes, Duolingo is a product-first company. When people are understanding Duolingo, the first thing they should understand is engineering a product, that is what the majority of our employees are doing. That is what we spend most of our money on. That is what we think about. That is what I think about. So that's the thing. Now that said, of course, we're not allergic to marketing. And in addition to that, because of the organic growth, we've actually built a pretty well-known and loved brand. And in particular because we have this owl mascot, I mean there's this green owl, that the app icon of Duolingo is a green owl, there's -- and it appears all over the place. So there's quite a love -- quite a bit of love for that mascot and for our brand. What that has allowed us to do, it has allowed us to basically be really efficient with marketing. So for example, on our TikTok, we didn't have a TikTok channel 6 months ago. We decided to start TikTok, and it has been extremely successful. We're now basically a case study of how to do TikTok. I mean there's all these news articles about how Duolingo is the only company that understands TikTok. I said I don't actually believe that's true, but we do understand TikTok. And I think a lot of it has to do with the fact that our brand and our mascot is so lovable. So if you go see our TikTok videos, they're basically -- it's somebody wearing the costume of our mascot, it's like this giant owl that's doing silly stuff. That has made it -- so our videos have been viewed by like more than 100 million people. And it's all entirely -- we haven't spent a single cent on boosting these videos or anything. This is all entirely organic. And I think it just has a lot to do with the power of our brand. Now when you asked about us spending money, we're not allergic to spending money on marketing. But when you're growing new products, my belief is that -- if you start spending on marketing too early, you get fooled into thinking that your product is good just because a lot of people are using it. But mainly people are just using it because you're spending $20 million on marketing somewhere, and that's why people are using it. So this is -- when you're early on, I am a big believer that you should let your products grow organically. With our Duolingo app, which is not just getting started, at this point, we are spending some amount of money in marketing. But the types of places where we spend money on are usually, for example, in underpenetrated markets where we would spend -- so an example is last year, we spent something like, I don't know, like about $1 million in Japan on mainly influencers to kind of get the flywheel going. And it really worked pretty well. So what you will see is you'll see us spend amounts of money on -- in underpenetrated markets to get the flywheel going. That's kind of the main expenses for marketing.

Great. That's helpful. And just shifting gears back to payer penetration. You mentioned earlier, it's 5.5%. But then when you compare that to, say, like mobile gaming apps, which is around 2%, and then dating apps, which is low double-digit percentage, where do you envision Duolingo trending over time? And then maybe to provide some context, what features or product improvements would you say aided to the growth to this 5% today versus, say, to only 3% back in 2019?

Yes. So this is something that we have a good number of people working on. We see -- the way we grow the fraction of paying subscribers is by basically running hundreds of A/B tests to try to get more people to subscribe. And that number seems to be -- the fraction seems to be growing by about 1 to 1.5 percentage points per year. And it's all because we've just run hundreds of A/B tests to get more and more people to subscribe. So -- and by the way, these A/B tests fall into mainly 2 buckets. The first bucket is changes to how we merchandise. So our purchase page when we say -- show the offer to subscribe, what we say there ends up mattering a lot. So do we give you a free trial or not? Do we emphasize the monthly plan versus the yearly plan? What do we say are the benefits of the subscription? What do we say in the button? So we run hundreds of A/B tests, and it just keeps getting better and better. So that's one. The other big bucket is adding new features to the premium subscription that get more people to subscribe. Now there would be a third bucket, which is one we're not using, which is another way in which you can get people to switch from free to paid is by making your free tier really crappy. So you could be like, for example right now, we put one ad at the end of a lesson. We could put 17 ads at the end of a lesson. And you're like, well, you've got to subscribe to turn those off. We know that, that would get more people to subscribe, but that would be a short-term benefit. Because what that would do is it would make our free user experience crappy, and then that would stifle our growth. Because most of our growth comes from people telling their friends, and they won't tell their friends if they see 17 ads at the end of a lesson. So what we're not doing is degrading our free user experience. But the other 2 buckets, that's what we're doing. And we just have hundreds of these tests still planned. And people ask us, how high can you get it? You're at 5.5%. How high can you get it? It's impossible for us to know how high we can get. But we think a good comparable are the dating apps. They're, call it, 11%, 12%. We see no reason why we can't get there because the business model is pretty similar, which is you give most of it away for free, but there are some perks. You turn off ads for the premium experience. So we think we can get it somewhere around there. Now are we going to settle at 10%? Are we going to settle at 20%? I don't know, but my guess is those are a pretty good comparable. And then in terms of big changes that we've made, just every year, we have a couple -- the majority of the improvements, by the way, are usually small changes. But every year, we have a couple of big things. A couple that we're working on today, one is we have a family plan that -- it's rolled out, but it's kind of not very heavily advertised within the app. We think that's going to be pretty big. Another one that we're working on is a higher level -- a higher tier of subscription. This is similar to what Tinder did at some point. I mean Tinder had their Tinder Plus at some point. They added Tinder Gold, and that was a pretty major unlock. We are working on a higher tier of subscription. So that's something else that we're working on that we haven't quite launched, but that's -- we're working on that.

Got it. Great. And then, yes, you mentioned the family plan being tested. I think I have that within my app today. I haven't seen the higher-tier plans, so that's interesting. And then also, I think you're also testing with regional pricing. So anything you can comment there in terms of differences within pricing? How -- any surprises there? And how does that impact you in terms of how to implement it globally?

Yes. So regional pricing, that's another thing. I mean I like talking about regional pricing because it gives you an idea of how early we are in our monetization journey. When we IPO-ed, which was, I don't know, 4 months ago, something like that, we had the identical price in every single country in the world. Now we know that is dumb. As in like, of course, you can do better. You should not price your app the same in the United States as you do in, I don't know, Brazil. You should really not do that. But it's just something we hadn't gotten to, so now we're currently working on that. By now, we have a different price in 4 markets only. So there's 4 countries where we have differentiated the price, and we're working on other markets. By the way, you may think, well, how hard could this be? Why haven't you just changed the price? And the reason is because we really -- for everything we do, we test it, and we're running A/B tests on this. And for changing your price point, the A/B tests have to run for a while because you have to measure what it does to your renewal rates. And so you kind of have to wait around for people to renew to see what it does. So this is why it's not moving in like a couple of weeks, which is you think you could do that. But over the next -- you will see us over the next few months that we're going to settle on different prices in every country. In most countries, we're probably going to lower our price point because most countries in the world have lower acquisition power than the United States. So we're probably going to lower our -- and I think that will increase our revenue. However, I don't believe -- some people are like, holy c***, that's probably going to triple your revenue. I don't believe that. And in fact I know that won't happen, so it will be an increase, but it won't be a major increase. And I'll tell you why. It's because right now, we make a lot of our money -- well, I mean, the majority of our money comes from digital subscription. Our digital subscription is priced right for the United States, and it's priced right for probably Western Europe. It's probably overpriced for Latin America and stuff like that. We're going to lower the price. And then it's going to be priced right, call it, in Latin America going to be priced right. However, that by itself is not enough for high conversion in these countries because in the majority of these countries, just digital subscriptions are not quite as advanced as they are here in the U.S. So people are just not as used to having digital subscriptions as they are in the U.S. Many of these people don't have Netflix. They don't have Spotify, et cetera. So when we lower the price in these countries, you will see an increase. And what it is -- what I believe it is, is this is a necessary but not a sufficient condition for us to get high penetration of payers in these countries. So yes, this will be an improvement, but it won't be a crazy, like tripling revenues because you lowered prices in Latin America. That's unlikely to happen.

Got it. Okay, noted. And then in terms of in-game revenue, today it's still a very small percentage of total revenue, despite Duolingo being a gamification type of app. And then you compare that to traditional mobile gaming where in-game revenue is close to 90% of total, so I personally think it's a very large opportunity for you guys in the long term. So I guess, one, do you agree with that? And then secondly, how focused is the company at ramping up in-game revenue?

It's -- I completely agree with you. This is a big opportunity for us. We are focused on it. I would say that's our second priority in terms of monetization. Our first priority in terms of monetization is our subscription. We are a subscription business. That's -- the majority of our money comes from our subscription. We'll continue with that. But we recently started working on kind of this in-app purchases, à la carte in-app purchases. Today, 2%, give or take, of our revenue comes from in-app purchases, from à la carte in-app purchases. But that has gone up from 1% like, I don't know, 6 months ago. So this has doubled in the last few months. And I do believe there's a lot. And there's so many reasons for that. And like you said, I mean games are one, but also the dating apps, which we look to a lot, Tinder, for example, 30% of their revenue comes from à la carte purchases. So our belief is that there is -- it should be a lot higher than 2%, and we have a team behind that, and we're going to do that. And there are a couple of reasons. One is just we're very gamified, and we can probably do a good job with that. But secondly, there are a good number of geographies where we have a lot of users where they're just not used to paying for recurring subscriptions, but they are used to paying for kind of onetime deals. And so our belief is that we can do a good job with that. And so that is something -- so yes, it's something I agree with and something that we're working on.

Great. Okay. So shifting gears a bit to the Duolingo English Test. It's expected to grow more than 50% this year despite pretty tough comps from last year due to COVID. What is the long-term goal here with the assessment tests? And then how soon can we see these tests implemented within the core app itself and then to other languages?

Yes. It's a great question. It's a major initiative for us. So for people who are unaware of the Duolingo English Test, it's a separate product today. It is -- if you've grown up in the U.S., you have no reason to ever take an English test. But outside of the U.S., standardized English proficiency exams are a big deal. A lot of people have to take them for many reasons. So we made our own English proficiency exam. You can take it online. And the idea is you just go there, you take a test. And at the end, you get a score from 0 to 160 about -- that measures how good your English is. There's many reasons why you may have to take these tests or why people may have to take these tests. But for example -- but in total -- by the way, so single-digit billions of dollars are spent every year by people taking standardized English tests. And for example, reasons why you may have to do that, if you apply to come to a university here in the United States from another country, you have to take that standardized English proficiency exam. If you apply to get a work visa in the U.K., you have to take that standardized English proficiency exam. So there's kind of many reasons. And before us, there were a couple of large tests that made on the order of, call it, $1 billion of revenue per year. It's probably a little lower than that, but it's high hundreds of millions of dollars per year of revenue. One of them was called the TOEFL. It's still called the TOEFL. Another one is called IELTS. And both of these tests -- and there's other tests, too, but all of these tests are pretty similar. They may -- they cost about $250, and you have to go to a physical testing center to take them. We thought we could do a lot better by basically making our test online. So as opposed to having to go to a physical testing center, our test, you can take it entirely online. It's also cheaper. It's $50. The reason, by the way, is because we just thought $250 is a crazy price. Most of the people who are having to take these tests are in developing countries, and $250 is a month's salary. And also the testing centers are not in every city, so you have to travel. So we thought, okay, this is crazy. We can do a much better job, and we can do it online. So we launched our test about 5 years ago. It's grown a lot. It is now accepted. By the way, if you launch your own test and you want anybody to actually take your test -- because nobody takes tests for fun. You actually want that -- you needed the results of your test to be accepted by organizations around the world. So we decided to start with getting U.S. universities to accept our -- the results of our test for international student admissions. We've made a lot of progress. So by now, by volume, about 80% of U.S. universities by -- when you weigh it by volume of international applicants, about 80% of the U.S. universities accept our test. Most universities you've heard of accept our test. And so that's already happening. We're working with -- in Canada, we're doing now with Canadian universities. U.K., we're working with U.K. universities, so we're increasing that. And then we're also working with governments for visa applications. So that's growing a lot. And I think that business is going to be -- grow a lot. That business, by the way, through COVID was crazy growth. It went from -- we went from selling 17,000 tests in 2019 to selling 344,000 tests in 2020. This year, we're going to grow by about 50%, so we're going to sell north of 500,000 tests. So that's 50% over a really hard comp, which was last year, like you said. And I think that business is going to continue growing. Right now it's 10% of our revenue. I think that will continue growing very nicely. And I see no reason why with that test, we can't make hundreds of millions of dollars. So that, I believe. But the reason I'm so excited about this test -- and it's related to one of the things kind of in your question is the test will allow us to do what I want to do, which is to have Duolingo be the standard of proficiency when people talk about their language. So as opposed to right now, when somebody asks you how much French do you know, and the most common answer is like, took 4 years of French in high school, which doesn't actually answer the question and who knows what that means. I want people to say, I'm a Duolingo 65. And the way -- the strategy for us that we're going to get there is we're going to -- we have this kind of dual strategy. One is the test, which -- the test by itself, what it does -- by the way, right now, the test is only for English, but we're going to do it for other languages. But what the test does is it adds legitimacy to the score. Because, well, that score is good enough to get into MIT, so MIT accepts the results of our test, that test is good enough to get employed. Great. That adds legitimacy. On the other side is the app. What we're going to start doing with the app is -- you're not going to be able to take the test from the app, at least it's not intentioned. But we're going to start adding an estimated score in the app. So when you use the app, after you use them for a while, we're going to say, hey, if you were to take the test right now, you could get a 35, or you could get a 60 or whatever. And the combo of these 2, I think, will be really powerful because what the app will do is be able give a scale. So we'll have literally hundreds of millions of people who have an estimated score. And then the test will give that score legitimacy. And the combo, the hope is that it makes it so that we become the language proficiency standard. And if we're able to do that, that would be a really defensible position.

Got it. Yes, that will be incredible. Maybe just lastly, this topic has been one that's kind of come up in between your IPO today. China recently banned education-for-profit companies, including Duolingo's app from the App Store. Any updates there in terms of China and how large that opportunity could be over time?

Yes. China, China. So okay, so the first thing to say is there's a ban in China for something very specific, which is kind of these after-school programs, et cetera. We do not fall under that. But it also happens that around the same time, we were taken off the Chinese app stores. So what that means is that users who already have the app can still use us. And so we still have hundreds of thousands of users in China actively using us. But new users cannot download us. So that is the current situation in China. I'm not going to try to guess what the Chinese government is going to do, but our belief is that this is a temporary situation, and we're working with them to resolve this. Now in terms of kind of how -- what this means for us, China has always been -- it's obviously a very interesting market. It's a huge market. I mean this is true for all -- for almost anything. Like China is a huge market, and it's a huge market for this. It has never been a huge fraction of a market for us. So it was the second fastest-growing country before this happened. So it's growing very fast, but it was only about 1.5% of our total revenues. And after the ban, our revenues are not going -- necessarily going too much down because our current users are still there. It's just our growth got stifled. I still think China is a big opportunity, but I just don't know what will happen. And when we have a lot of investors who kind of ask us like what do you think China's going to do, the truth is I am not an expert in geopolitics. I don't know what China is going to do. But it's obviously a great opportunity, but not one that we -- not one that it would be intelligent for any U.S. company to bet their entire company on. So it's a nice -- it's kind of like -- it's a nice kind of option, like maybe it will be good.

Yes, call it option is a good way to put it. Great. Yes, unfortunately, I think we're out of time. So Luis, thank you so much for joining us today. And hopefully, we'll meet in person next year.

Yes, excellent. Thank you, Mario.

Okay.