This podcast episode is brought to you by Helix Sleep. Sleep is super important to me. In the last few years, I've come to conclude it is the end-all be-all. That all good things, good mood, good performance, good everything seem to stem from good sleep. So, I've tried a lot to optimize it. I've tried pills and potions, all sorts of different mattresses, you name it. And for the last few years, I've been sleeping on a Helix Midnight Luxe mattress. I also have one in the guest bedroom and feedback from friends has always been fantastic. It's something that they comment on. Helix Sleep has a quiz takes about two minutes to complete that matches your body type and sleep preferences to the perfect mattress for you. With Helix, there's a specific mattress for each and every body. That is your body, also your taste. So, let's say you sleep on your side and like a super soft bed, no problem. Or if you're a back sleeper who likes a mattress that's as firm as a rock, they've got a mattress for you, too. Helix was selected as the number one best overall mattress pick of 2020 by GQ Magazine, Wired, Apartment Therapy, and many others. Just go to helixsleep.com/tim, take their 2-minute sleep quiz, and they'll match you to a customized mattress that will give you the best sleep of your life. They have a 10-year warranty, and you get to try it out for 100 nights risk-free. They'll even pick it up from you if you don't love it. And now, my dear listeners, Helix is offering up to $200 off of all mattress orders and two free pillows at helixsleep.com/tim. These are not cheap pillows either, so getting two for free is an upgraded deal. So, that's up to $200 off and two free pillows at helixsleep.com/tim. That's helixsleep.com/tim for up to $200 off. So check it out one more time. Helixsleep.com/tim.
This episode is brought to you by Athletic Greens. I get asked all the time what I would take if I could only take one supplement. The answer is invariably Athletic Greens. I view it as all-in-one nutritional insurance. I recommended it in fact in The 4-Hour Body more than 10 years ago, and I did not get paid to do so. With approximately 75 vitamins, minerals, and whole food sourced ingredients, you'd be very hard-pressed to find a more nutrient-dense and comprehensive formula on the market. It has multivitamins, multi-mineral greens complex, probiotics, and prebiotics for gut health, an immunity formula, digestive enzymes, adaptogens, and much more. I usually take it once or twice a day just to make sure I've covered my bases if I miss anything I'm not aware of. Of course, I focus on nutrient-dense meals to begin with. That's the basis. But Athletic Greens makes it easy to get a lot of nutrition when whole foods aren't readily available from travel packets. I always have them in my bag when I'm zipping around. Right now, Athletic Greens is giving my audience a special offer on top of their all-in-one formula, which is a free vitamin D supplement and five free travel packs with your first subscription purchase. Many of us are deficient in vitamin D. I found that true for myself, which is usually produced in our bodies from sun exposure. So adding a vitamin D supplement to your daily routine is a great option for additional immune support. Support your immunity, gut health, and energy by visiting athleticgreens.com/tim. You'll receive up to a year supply of vitamin D and five free travel packs with your subscription. Again, that's athleticgreens.com/tim.
Optimal minimal. At this altitude, I can run flat out for a half mile before my hands start shaking.
Can I answer your personal question? Now would seem a perfect time. I'm a cybernetic organism living over a metal endoskeleton.
Hello boys and girls, ladies and germs. This is Tim Ferriss. Welcome to another episode of the Tim Ferriss Show. My guest today is Luis Von Ahn. You can find him on Twitter at Luis, L-U-I-S, V-O-N, Ahn. Luis is an entrepreneur and consulting professor at Carnegie Mellon University who is considered one of the pioneers of crowdsourcing. He is known for co-inventing CAPTCHAs, being a MacArthur fellow, and selling two companies to Google in his 20s. He's currently the co-founder and CEO of Duolingo, a language learning platform created to bring free language education to the world. With more than 500 million users, it is now the most popular language learning platform, and the most downloaded education app in the world. Luis has been named one of the Brilliant 10 by Popular Science, one of the 50 Best Brains in Science by Discover, one of the Innovators Under 35 by MIT Technology Review, and one of the 100 Most Creative People in Business by Fast Company. Luis also won the 2018 Lemelson-MIT Prize, the largest cash prize for invention in the United States. Luis, it has been a while since we spoke. Last, it was 2016 when we had a conversation on this podcast. Welcome back. It's nice to see you.
Thank you for having me, Tim. And I thought we could start with just a few bits of trivia, a little bit of history, and why don't we start with why the Duolingo owl is green. Would you mind just explaining that for people?
Hopefully, most people have seen the Duolingo owl. It's our mascot. First of all, it's an owl because in most Western countries, owls represent knowledge. Turns out this is just not true at all in Asia. And now that we've been expanding to Asia, people ask us like, 'Why the hell are you using an owl? Owls are just kind of vicious animals.' But okay. It's an owl and it's green and it is a ridiculous reason why it's green. When we were getting started, my co-founder Severin, who's an amazing guy, his last name is Hacker. Severin Hacker.
I still think it's made up. He swears it's not. This is a computer science person whose last name is Hacker. But he, you know, we were getting started and we had a design team that was working on our logo and everything. And we thought we wanted a mascot. We thought it was important to have a mascot because well it would kind of help people learn and it would make our brand iconic etc. And that turned out to be true and we came up with the idea of an owl and then you know we were just talking about things and then Severin was not very involved in the design process at all and at some point the designers asked him, well is there anything you care about in terms of design? And he said, 'Look, there's only one thing I care about, and it is that I hate the color green.' And so, we thought it would be funny if he was the co-founder of a company where most of our colors were green. And he's had to live with that for, you know, 10 years.
All right. And we will try to not replicate too much of the first conversation, but I think a little connective tissue and context for folks would be helpful. So, I'm going to try to do a quick recap of a few points and I want you to correct anything that I get wrong and certainly add in any color that you think would be helpful. So, you were born and raised in Guatemala and you were raised by two parents, both medical doctors. You ended up, I think it started with receiving a Commodore 64, much to your disappointment instead of a Nintendo, becoming involved with taking things apart and learning computer science. And flash forward, you and I'm skipping a lot of steps here, but had developed a technology and a company, reCAPTCHA, which after about a year and a half sold to Google. Most people have filled out a CAPTCHA or reCAPTCHA to prove they are not a robot or a bot. And a year and a half later, this Severin Hacker who you've already mentioned, who is then a PhD student of yours, was looking for a PhD project. And that led to a conversation including the observation that number one, language, or I should say education was important to both of you and that in your home country of Guatemala education was in a sense an obstacle to overcoming socioeconomic barriers and to elevating your life because of the cost involved. So the rich were able to educate themselves and sort of perpetuate that cycle of wealth whereas the poor were not. And there were many many different facets of this which led then to what would later become Duolingo. Is that roughly accurate?
That's exactly right. That's exactly right. We can wrap up now. We're done.
We're done. We're done. All right. And you and I, I don't know when you and I, I can't recall offhand when we first connected. I want to say it was somewhere between 2010 and 2012. It must have been.
I think it was around 2010. It was around 2010 or 2011.
Yeah. So, right in that range and early days with respect to Duolingo and the last time we spoke you had, this was in 2016, last time we spoke on the podcast, roughly 60 employees. So, could you just give us a then and now comparison of Duolingo then, Duolingo now, what the biggest differences are?
We've grown a lot as a company, of course. We now have about 600 employees. So of course 10x'd in employees. Back then we were a privately held company VC funded. We are now a publicly traded company. That was a big milestone for us. Another big thing that changed between 2016 and now is we really figured out how to monetize Duolingo. When we launched Duolingo, the thing we cared the most about is that it was a free way to learn languages. And at first we just weren't concerned too much about figuring out how to make money. And the way we supported ourselves was through venture capital. We just kept on raising more and more venture capital and at the time the prevailing wisdom was hey don't worry too much about making money just grow your users and this is what we did and the way we grew our users by the way is not through marketing because since we weren't making any money we thought it was kind of a waste of our money to spend to acquire users if we weren't making money on the other side. So the way we grew our users was just by making the product better and better and it grew through word of mouth.
And I think one thing that happened is around 2015, 2016 we had raised a bunch of venture capital. I mean probably around a hundred million dollars total at the time and we still weren't monetizing at all but we had grown to a pretty good amount of users, active users. We raised a round of funding from at the time it was called Google Capital and now it's called Capital G. And the partner that was on our board, Laela Stury, took me out to a bar and she said, 'Listen, you've been raising a lot of capital and your valuation keeps going up and up. At the time, I think we're valued at something like half a billion dollars or something, but right now you just raised money from Google and let me tell you, there is no bigger fool. Like, you're not going to find another bigger fool to raise money at a higher valuation with no revenue. You got to figure out how to make money.' She gave me enough drinks that convinced me to promise her that next time we spoke in six months, I would have figured out how to make money.
And so I went back and like, 'All right, everybody, we got to figure out how to make money.' The easiest way to make money when you're teaching somebody is just to charge them to learn. That's kind of how education monetizes. It was very important for us that we didn't do that because the mission of our company was really related to giving access to education to everybody for free. So, we're a little constrained here. In the end, we ended up finding a monetization model that worked out really well, which is this premium model that is similar to what the dating apps do or what Spotify does, which is you don't have to pay, but you may have to see some ads at the end of a lesson. We make money from the ads and then if you don't like the ads, you can pay us to subscribe to a premium version of Duolingo, and then we turn off the ads and we give you a few extra perks. And that combo worked out really well.
I mean well enough that that is basically what we used to IPO. It has grown to hundreds of millions of dollars of revenue and so that worked out really well and we're also very proud that because of that Duolingo, it used to be that Duolingo was the most downloaded app in the education category, it still is but now it's also the highest top grossing app in the education category. So we make more money than all other education apps and we're very proud that you can still use Duolingo entirely for free. So even though most other education apps you have to pay to use them, in fact, something like 97% of our active users use Duolingo for free. Still, we make more money than all other education apps. So we're very proud of that. The monetization was a big shift in our company.
I have questions about the monetization and then I want to come back to Laela and talk about org chart. Maybe another conversation, maybe fewer drinks.
I'm happy to talk. Also, Laela is one of my favorite people. Laela is awesome.
Yeah. So, we're going to come back to Laela and give her more credit. Do you remember what types of drinks she was feeding you? Do you have any idea? Or did the memory go out the window?
You know, I don't know that. And honestly I don't know. It was at a bar. It was at a bar near the Duolingo office. It's about a block away from the Duolingo office. Kelly's Bar.
Kelly's Bar. All right. So, the scene of the crime. So, we're going to come back. We're going to come back to that. I wanted to return to monetization. And I'll just mention to folks as foreshadowing. The reason I'm interested in talking about the org chart is because again this will test my memory of the first conversation but as I recall when you sold reCAPTCHA to Google you had maybe six, seven or eight employees something like that. Is that rough?
Around 10. Yeah. Around 10 employees, about 10.
So let's round it to 10. And then for a while, I want to say maybe into the 30 to 40 employee range, you had a very flat organizational structure, so to speak, effectively a lack of hierarchy at Duolingo. And although you had a lot of experience with computer science certainly and product development and thinking about the technical side of things, building a company of 600 people was not one of them at that point in time.
I had no previous knowledge of how to do this.
Yeah. Right. So that's why I want to come back to the org chart. But first on the monetization side, what are some of the things that you tried that didn't quite work and why did they not work just in terms of whether primary or ancillary models because that I recall as someone who is one of the pioneers of crowdsourcing you certainly have come up with many different ideas for how you might approach this.
The first thing we tried and this is before the Laela talk and it was just very early on we tried this and it was a really clever idea but ultimately I just don't think it was a very good way to monetize. Very early on when we had started Duolingo what we said we would do is look there were all these people that were learning a language on Duolingo for free. We thought rather than giving them ads or charging them we would get them to do some work for us in exchange of the learning. And so the idea was the following. In fact we built this and it worked. The idea was the following. Imagine you were a Spanish speaker who was learning English on Duolingo. You would learn for free, etc. At the end of your lesson, we would say, 'Hey, you just practiced food words. Do you want to practice with a real-world news article about food that is in English? You're learning English. And to practice, can you help us translate that into your native language of Spanish?' And you know, we would give you like a paragraph of that or something. And then we would give many people the same paragraph and try to get them to translate it. And it turned out that actually got really good translations.
So what ended up happening is we signed a contract with CNN where CNN would send us all their news in English. Then our users to practice their English would translate those articles into Spanish. And then, you know, we kind of combined all of the different translations into one translation. And we had all this kind of multi-step process to clean up the translation, etc. and then we would send back a translated version to CNN into Spanish that they could post in their Spanish site and they would pay us for that. So we signed a contract with CNN. We signed a contract with BuzzFeed and this worked and it sounds very clever. The problem is as a monetization technique, first of all, translation is a bad business to be in. Computers are getting better and better at it. And even without computers, it's just kind of a race to the bottom in terms of price because you can always just find somebody else to do it cheaper for you. So that was kind of one thing. And then we just ended up realizing this was not an easy way to make too much money.
The other thing is we started realizing because that's where the money was coming from. We started realizing we were spending a lot of our thinking time and a lot of our development time on making sure the translations were accurate. Whereas to begin with what we wanted to do was make an education company and we wanted to spend most of our time teaching better rather than cleaning up translations. And so at some point very early in the company we weren't making very, we were making I mean like thousands of dollars, not a lot of money. We decided we took this hard decision. This must have been like 2013 or something. We took the hard decision to stop that and we're like, we're not going to do that and we're not going to worry about making money and we're just going to do venture capital for a while. Like don't worry about it. So that was one.
The other one is after the Laela conversation when we needed to really figure out how to make a lot of money. Now the thing is when you have a half a billion dollar valuation, you got to make a lot of money. Like when you're making zero, you can't make it so that you're only going to make $100,000 a year. Like for a half a billion to justify a half a billion dollar valuation, you have to make it. You can't turn it into a lifestyle business for yourself.
No, no, you got to be making, you know, at least tens of millions, if not hundreds of millions of dollars in revenue to have that.
So, we thought, okay, the first thing we thought, by the way, when Laela said, 'You have to monetize,' I went back to the company. And, you know, at the time, we had maybe 60, 70 employees. And I said, 'Okay, well, we had to figure out how to make money.' And the first question everybody had was like, 'Why?' I was like, 'Well, companies make money. That's how they support themselves.' And they're like, 'Well, we've been fine without that.' Then I had to explain you just cannot raise venture capital forever. Then it's like okay fine we understand we have to make money so how are we going to do it? And then I would say well you know maybe we can put an ad at the end of every lesson like no no no ads are evil. Don't put an ad. Okay have a subscription. No you can't charge people. You can't do that. And so it's like the whole company was basically against monetizing.
And it took us about six months to finally convince everybody that monetization was not a bad thing and that in reality what it would allow is if we could make a lot of money we could use that to invest it and to teach better. And people actually finally understood. But it took a while. And in that time at some point one of the first few things that changed was people were like okay you know what we could do ads. We could do ads but they can't be the crappy ads that are all out there. Let's do a clever idea. And the clever idea was and it's a great idea. We just never made it work. And I don't know if at some point I do want to go back to it. Can we show ads in the language that the people are learning? So if you're learning Spanish, could we show you some ads in Spanish? And I think that's a really good idea because it's kind of like a win-win situation. Like the person wants to actually hear the ad or see the ad, they practice, and then we make money off of it.
There's one big problem for that, which is that usually companies their budget doesn't work that way. So, for example, in the US and you're advertising to American consumers, Coca-Cola, I'm sure, has an ad in German, but they just do not want to show the German ad. That's just not what they want to show because they're like, 'No, no, no, no. In German, we have a very different narrative. That's just not what we want to do. We want to show the American ad.' And we're like, 'Ah, can we show the German ad?' That just we never got it to work. I thought it was a good idea. Never got it to work. In the end, now we reverted to showing standard app ads that we show that are, you know, what you're supposed to get if you are in the US, you get the English ads.
May I just because I'm overcaffeinated, just throw something out there really quickly, which is
And I'm sure with your technical abilities, you could figure this out or the advertiser could just apply the raw text of course for closed captioning that they would use otherwise, but you could show the translated subtitles in the target language, say of German.
Yeah. And in fact, that is a good idea and we probably should at some point. There's one other complication with this, which is right now the ads we show, they're programmatic ads that come from like Google and Facebook. We show Google and Facebook ads and Google and Facebook are the ones who serve us the ads. If we wanted to do something like this, we would have to have our own ad sales force. So, we would have to go and contact companies and ask them to do that. And that's doable and probably you can make more money per ad impression if you do that. But it's the pain in the ass of having basically an ad sales team because our monetization particularly subscription has done so well. We haven't felt the need to do that. I mean at some point maybe we'll do it but we just haven't felt the need to do that.
So I want to give you credit where credit is due on I'm sure there are many things I could give you credit for but one is a relentless focus and ability to refocus. And I want to mention an article for folks it's very short that they might find interesting. It's called 'The Top Idea in Your Mind.' It's by Paul Graham, which relates back to the point you made of realizing that you're not in the business of tidying up the last 5% of translations that you have sourced from your users and in fact you're an education company first and foremost. So being able to say no to a good even a profitable activity in order to remain focused just as you would not want to necessarily build your internal sales team right because it would just detract from the focus and resources. I want to return with a question about the months it took to convince the team, the employees, to get behind monetization of some form or another. How did you do that? And how much of it was like here is the mission? Because presumably they care about the mission of the company, but they also care about the value of their equity and want liquidity at some point. So you could just say like, 'Hey guys, if we keep raising venture capital, like there might be secondary sales, but at some point both the investors and employees will want to be able to cash in on at least some of the value of their equity. That's part of the reason you came to this company rather than going to work for someone else.' What was the conversation or the process of persuasion?
You know, it's interesting. The first few employees that we hired at Duolingo, by the way, I think a lot of it has to do with the fact that we have a really good company culture, but also that we're in Pittsburgh, our employee churn is very low. Employees stay for a very long time at Duolingo. So, the majority of people that started at Duolingo 10 years ago, they're still around.
And the thing about these people was the first few employees that we hired were people from Carnegie Mellon University because I was a professor there. And I basically just knew who the best students were and I'm like, do you want to come work for us? All these people at the time had offers for the usual suspects, Google and Facebook and the usual suspects and at the time we had just raised our Series A but we could not compete salary wise with Google or Facebook or anything like that. Now we can compete, our salaries are comparable to Google or Facebook but back then we just couldn't. So the people that took a job with Duolingo were taking a major pay cut, sometimes like a 50% pay cut, like we're paying half the salary that they would have gotten at Google and they had offers at Google and they took the job and the reason they took the job was they really cared about the mission.
So you know I like to tell Severin and I care about the mission. These first few employees were zealots. They were like, 'Look, this is the main thing that matters to me.' And so these people for years have been kind of the guardians of the culture. And so Duolingo was a very mission-driven company. And of course it has to do with Severin, but I think it also has to do with this first few employees who just really care and they're still around. And so when we were starting to monetize, the discussion was really never about like, hey, you could make a lot of money off of your shares or whatever. That just didn't resonate all that well. What ended up resonating the most was, 'Hey, look, right now we have about 60 employees. We could hire way more employees if we can make a lot of money.' And the majority of them can be working on improving how well we teach.
And that really resonated and it became true. I mean, today most of our employees are actually working on making our product better in terms of how well we teach, how engaging it is. And it is because we make hundreds of millions of dollars per year. We use that to fuel the mission. So, I think there was just a shift in thinking that really worked and we believe it. I mean, the company now believes that.
I'm glad I asked. I'm glad I asked. These stories matter. And I don't know the answers, by the way, people listening. I mean, Luis has been busy to put it mildly. We don't always catch up very frequently. So, this is an excuse for us to catch up as well. Let me ask you a question that I would ask if we were just having drinks over dinner. I would not ask you to make any promises about monetization to me. But the broader question since you mentioned computers getting better at translation. So this doesn't directly relate to Duolingo I don't think but it might in some indirect way and for people who don't know a bit of the history if we look back at say reCAPTCHA part of what fascinated me about reCAPTCHA is like all right if you and I think you did run the kind of basic arithmetic on how many human hours were going into verifying that each individual is not in fact a bot began to wonder is there a way to put these human hours to work for some benevolent purpose or a positive purpose and at least one facet of that ended up being helping to digitize books. So using human eyes and then just kind of law of large numbers to ultimately get a good transcription of digital books that machines were having a hard time with. What are some of the most interesting things or interesting developments that you're seeing with respect to crowdsourcing versus just raw computing power or new technological pure kind of machine-driven technologies?
Well, you know, computers are just getting a lot better at things compared to 10 years ago, 20 years ago. AI is just a lot better. For example, at this point, it used to be that CAPTCHAs were mainly these distorted characters. At this point, computers are almost as good, if not better, than humans at reading these distorted characters. So, showing some distorted characters is not a good test to distinguish humans from computers anymore. If you've seen with reCAPTCHA lately, what they show you is these kind of images of like find the traffic light or find the stop sign or find what they're doing there is they're crowdsourcing things for like mapping software or self-driving cars where self-driving cars may take a lot of pictures for things and they may see 90%, I don't know the exact fraction but they may see 90% of traffic lights and realize that it's a traffic light but for some of them they just can't quite tell. And so what reCAPTCHA is doing is it's taking some of the things where the computer's not super sure about and it's getting people to kind of tell them that's helping to improve the technology. That's something, you know, I really like that application. I think it's worked out pretty well. So that's helping to improve technology.
Now, in case of language translation, that's got pretty good. I mean, if you try it and it's kind of good enough and if you're going to go to Germany or something, it's kind of good enough to like if you don't know any German to just kind of, you know, you speak to it and it speaks out something and you sort of can communicate. You know, people have asked us about Duolingo like, well, are people still going to need to learn a language given that translation's getting better and better? And it's one of those where like I don't think this is going to affect us at all. There are these use cases like if you're going to go on a business trip for two days and you really have no interest in learning a language, yeah, sure, use Google Translate or use whatever it is you need to use. But there's a couple of things. First of all, no matter how perfect the translation is, there's this thing that happens that there's always going to be a delay. Even if it's like supercomputing power, etc., because word order changes.
So German is a good example. In German, usually the verb is at the end. So, a German sentence, the way you would say it kind of in English is like, I did something yesterday. I did it really fast. I did it with my friend and what I did was run. And so, you kind of have to wait until the verb comes out to start saying it in English because in English you would say, I ran with my friend yesterday really fast. So, there's always this kind of like pause that makes conversations pretty awkward. Like you just can't really live your life that way. And the majority of language learning is either to learn English and actually say move to the US or actually learn English for business reasons etc. You actually want to learn English. And then the other big chunk is hobby and young people are still going to learn as a hobby because it's the same reason why we learn all kinds of things that computers could do really well like music. So I just don't think that's going to affect us as a company in terms of language learning.
Yeah, I don't either. I mean, I have been incredibly impressed specifically with Google Translate with respect to, in my case, East Asian languages because I do still communicate a fair amount in Japanese and to a lesser extent Mandarin, but
The translation specifically in Japanese has improved tremendously from even...
Yeah. It's incredible. You want to talk about the sentence construction and word order changing. Certainly in Japanese that is true because it's just for people who are interested in this kind of thing but don't know very much about it. If you have say I eat the apple in English it's subject verb object. In Japanese generally speaking it's going to be subject object verb. So I the apple eat. I to the school go, that kind of thing. And it's improved a lot but there is that delay as you mentioned.
And if I look at the, I'm looking at duolingo.com/courses right now and this may not be the best representation of use on the platform, but I'm curious to know who the primary users are in the pie chart if you're just going to break down the top few categories and what they're learning. And I was surprised to look at courses for English speakers. It looks like Spanish is number one. Okay, not too surprised there. Then French. Okay, interesting. Very interesting because yeah, French is certainly spoken by the sort of French diaspora and other places, not just in France, but
It's a result of colonialism.
Right? Exactly. So that's what I figured.
French is a result of colonialism.
Colonialism. But then you have Japanese, Korean, German, Italian, Hindi, Chinese. So Hindi, Chinese, I think specifically Chinese, I would have expected to be a little larger. And I'm astonished that Japanese and Korean, especially Korean because I could even chalk up Japanese to like nerdy tech folks who are Japanophiles and love Japan animation and so on. I could chalk up some of that to that, which may not have any accuracy to it at all. But then Korea,
Okay. All right. Great. And then Korean though, I was like, why Korean? Why is Korean number one?
K-pop. K-pop. K-pop. K-pop.
K-pop is humongous. I mean, K-pop and also Korean TV shows, you know, on Netflix. There's just been Korean entertainment. Squid...
Is really gaining a lot. Squid Game.
Yeah. The Squid Game. That movie. What was it called? The one that won the Oscar.
Oh, I know the one you're talking about.
Parasite. No. So, stuff like that. It's just Korean influence just keeps growing and growing. And so it is interesting. More people in Duolingo are learning Japanese or Korean than Chinese. And Chinese is an interesting one. There's the size of the language in terms of native speakers.
And in this case, we're talking about Mandarin, right?
Mandarin. Mandarin Chinese, you know, it's the largest language in terms of native speakers. And then there's a different number which is the number of people who want to learn it. We have a pretty good idea of the people who want to learn it because when we offer it on Duolingo, we can see how many people actually learn it. There's some that are flipped. I mean, Mandarin Chinese is flipped in that. There's a lot of people who speak it natively. Not that many people are learning it. So, I don't know exactly off the top of my head, but about 1% of our active users are learning Chinese, which I would have expected it to be higher. And in particular, part of the reason why I would have expected it to be higher is because the circles that we move in, I assume they're similar to the circles you move in, it's kind of educated people, tech people, they always have their kids learning Chinese.
Yeah. Yeah. This is the thing. You go to Palo Alto, it's like every other parent has their kids in bilingual nursery school or whatever. Yeah.
So you think the whole world's learning Chinese, but this is actually not the case. I think if you go anywhere in middle America, you know, the school only offers Spanish, right? Like that's just to begin with. And so this is and I think there's a number of reasons, but I think probably the biggest reason for Chinese that not as many people are learning it is it's just a very hard language to learn.
And I think people know it and they're like, 'Ah, it's hard.'
So in the US, our top languages are exactly as you said, Spanish, which makes sense. We are very close to Mexico and Latin America and it's just a very large language. French, which I think is a beautiful language and everything, but I think a lot of it, that one is the opposite of Chinese in that the number of people learning it versus the number of people who are native speakers, it's just a lot of interest for French given the number of native speakers. And then Chinese and Korean are pretty high up and Italian and German.
Now, you were asking about the types of users in Duolingo. I think we have two big buckets of users. In my head I classify them as kind of two big buckets. One big bucket which is about call it 50 to 60% of our users is people learning English. These are people who are usually not in the US. I mean some of them are in the US but they're usually not in the US. They're usually in a non-English-speaking country and they're learning English. The reason they're learning English is they actually want to learn English and the reason is because they want to get ahead in life. So either they want to get into a better educational institution or they just want to get a better job. The thing about English that is so magical is that anywhere you live if you speak English just by speaking English you can make more money. It's direct. So for example if you used to be a waiter now you can be a waiter at a hotel, makes more money. If you used to be a personal assistant now you can be a personal assistant for somebody in a multinational corporation or something. So it's just knowing English just gives you better opportunities in life and it's by far the most learned language in the world certainly by far the most learned language on Duolingo and those people are very committed.
And then the other big bucket of people are people who are English speakers already who are learning another language. So think of this as a person in the US who's learning Spanish or French etc. They're pretty different. It's more of a hobby. Obviously I'm generalizing here. I mean, there are some people obviously who need to move to Spain and they're going to learn Spanish for that. But for the majority, it's kind of more of a hobby that they're learning the Japanese or Spanish or whatever. The funny thing about these people is when you ask them, 'What would you do if Duolingo went away?' The most common answer is, 'I would spend more time on Instagram.' They're not that committed. They're like, 'Ah, you know what? I like it. The reason I'm learning Spanish on Duolingo is because I used to play a lot of Clash Royale or a lot of some game and it was a complete waste of my time. Now I'm using Duolingo. It's pretty fun and at the very least at the end of the day I'm learning some Spanish.' They're an interesting group of people.
The other thing to say is in the US 80% of our users were not learning a language before Duolingo. So we're completely growing the market. That is not true in a country like Brazil where people are mainly learning English. These people were learning English before Duolingo. They'll continue learning English even if Duolingo were to go away. These are the two big groups of people I'd say.
Just a quick thanks to one of our sponsors and we'll be right back to the show. This episode is brought to you by LinkedIn Jobs. It's summer 2022 and many small business owners are busier than ever. Finding the right candidate to join your business can be the leverage you need to be able to ignore the trivial. Delegate some of the urgent stuff and focus on longer-term strategic decisions. The big picture. LinkedIn Jobs makes it easier to grow your team by helping you find the people you want to interview faster and for free. Create a free job post in minutes on LinkedIn Jobs to reach your network and beyond to the world's largest professional network of more than 800 million people. Then add your job and the purple hashtag hiring frame to your LinkedIn profile to spread the word that you're hiring so your network can help you find the right people. Simple tools like screening questions make it easy to focus on candidates with just the right skills and experience so you can quickly prioritize who you'd like to interview and hire. It's why small businesses rate LinkedIn Jobs number one in delivering quality hires versus leading competitors. LinkedIn Jobs helps you find the candidates you want to talk to faster. Did you know every week nearly 40 million job seekers visit LinkedIn? Post your job for free at linkedin.com/tim. That's linkedin.com/tim to post your job for free. Terms and conditions apply.
I'd love to just get a fact check if you know here. I mean, maybe the math is so easy that it's almost self-evident, but I recall someone saying to me, which kind of makes sense, although I haven't done the math on it, that there are more people who speak English or are learning English in China as Chinese nationals than all of the native speakers of English in the world combined.
That's probably close to true. Yeah, I would imagine that. That might be something like India as well. I mean, although you do have the British history, so maybe there are
It's probably close to native speakers of English. Let's think about that. There's 300 million-ish in the US, but then there's the UK, too.
Yeah. And, you know, Australia, Canada, etc. They probably combine that another 100 million-ish.
So, very rough numbers here, but we're talking about 400 million-ish.
Yeah. It seems like China's probably not quite there.
Well, it could, but there's a big butt here. They may all be claiming to be learning English. And in reality, they may be learning English, but the learning outcomes may not be so good. So, it's like, yes, they're learning English, but maybe they know how to say hello and good morning and where's the bathroom? But that's about it. So, it's not clear that they're really good. I do not believe that in China there are 400 million really, really fluent native speakers. I don't believe that to be true.
I don't believe that either. And anyone who travels widely in China, I think, would probably come to that conclusion.
That's just not the case. Yeah.
So, let's, as I promised earlier, come back because I'd love to talk about some of your lessons learned, key decisions in growing a company of this size for the first time. And there are many ways we can tackle this. We'll talk about the personal. So just in terms of kind of personal management, work-life balance if that exists, lessons learned, but let's also come back to the org chart and Laela. Could you describe how that was done, what it ended up looking like, what were the implications and just kind of walk us through that. And I remember in our first conversation I told the story of maybe not the best example since it ultimately has been superseded by Netflix and superior technology but Blockbuster back in the day when Blockbuster as a company was about to become a blockbuster brought in I believe someone from McDonald's, an exec to specifically help them with org chart design. I've never designed an org chart. So could you just describe what happened and where you landed with it?
I mean, we've gone through a bunch of different org charts. When we were from zero to call it 30 employees, our org chart was very flat as in I was managing everybody. That was that. And I actually think, you know, maybe I took it a little too far, but I actually think for the first 0 to n where n is around maybe 20 to 30 people, the best thing you can do as a CEO is be a micromanager. I actually believe that when you are such a small company, usually you don't quite yet have product market fit. You haven't really figured everything out. You have one goal, get to product market fit.
I don't think you should be in the business of coaching people for this or that. No, no. I think you just micromanage people to get to product-market fit. I actually believe that at some point it really shifts. Even if you love micromanaging, which I love micromanaging, but I've learned not to do that anymore. Even if you love it, at some point this just becomes you just can't do it as well. And it is in your best interest to start actually developing people. You know, that shift should happen maybe at around 20, I don't know the exact number, that is when you should really start having kind of a couple of managers. I think splitting things up into teams, what we did then, it's not like we had a pretty good idea of exactly what our org chart should look like, but what we did was we hired our first manager that wasn't me. It's a woman that—this was, I don't know, seven years ago, something like that. It's a woman that's still with us. She's now the head of all of engineering. She used to be a director of engineering at Google. I knew her because after selling reCAPTCHA I spent some time at Google. I knew her. I really liked her. Her name is Natalie Glance. She's amazing. And she took a much smaller job at Duolingo. She was managing, you know, a larger job that she had, and she took a smaller job at Duolingo. I think she really believed in us, and she helped us. From her is where I learned how to manage people really much better. She actually knew how to manage people, and she really helped us kind of start having a structure of, well, we have some managers. That really made a big difference, starting to have teams. Then the thing that we did after that is we discovered this idea of metrics-based teams, which to this day we use, and I think has been a really, really good thing. I think this is not the common thing, although some companies do do it. It's not the common thing. So the standard thing that you would have in a company, for example, at Duolingo, we have—the app has a bunch of different features. One of the features could be like we have a leaderboard. For example, we have a leaderboard, that's a feature. In many companies, commonly, the normal thing to do is that you have a leaderboard team, a team that owns that big feature. So just kind of you split it up by feature, these are feature-based teams. We do not do feature-based teams. We do metrics-based teams. So we don't have a team that owns the leaderboard. Instead, we have a bunch of teams that own each a single metric. So for example, a metric that we have is 'time spent learning.' What they own is the number of minutes per day that the average user uses Duolingo for. And it turns out that changes to the leaderboard can increase or decrease time spent learning if you do kind of the right or the wrong thing. So that team messes with the leaderboard a lot, but they don't own it. The only thing they do is they have this one metric, and every quarter it has to increase. And so they just work on increasing this one metric and they run hundreds of A/B tests to increase this metric. We discovered that, you know, soon after Natalie showed up, we discovered these metric-based teams. And we started at first, our first metrics-based team was a retention team, which is all you have to do is make sure that users come back every day. And that team, you know, has done all kinds of things, really optimized the streak on Duolingo. So this notion of a streak that people have. You have people who have used it what, every day for seven years. I was looking at the streak.
Oh, yeah. Yeah. Yeah. Yeah.
The streak is crazy. Um, we have, well, the number is a little larger, but now that we're a public company we can't reveal numbers. As the latest number we've revealed is that we have 1.5 million daily active users that have a streak longer than a year, meaning they have not missed a single day in the last year or longer. That's a very powerful mechanic. The retention team has worked on that. So basically what we did is we discovered these metric-based teams, and what we do now at Duolingo is when we care about a metric that we want to optimize—so a metric could be daily revenue or, you know, whatever it is—we form a team around that metric. That has worked out pretty well. So that's kind of one big shift with us that happened. So that worked out pretty well. At some point we had like 50 teams, I maybe a little less, maybe like 40 teams. That was started to become—we use the term 'goat rodeo'—it started to become this craziness. And so what we did is we decided to pull the teams together that were similar to each other into this thing called 'areas.' So now, for example, we have a monetization area. Inside the monetization area, we have a team that owns ad revenue per day. We have a team that owns subscription revenue per day. And then we have another area called the growth area, where it's just growing our active users. We have a team that owns time spent learning. We have a team that owns retention. We have another team that owns new user retention, etc. And so we split up into areas, and that's how we're split up now. And what has been really good is, and the shift for me has been, it went from micromanager to kind of learning how to manage to now what I do is, you know, I'm a manager of managers. Well, I'm a manager of managers of managers of managers. But at some point that kind of doesn't matter that much. You're just a manager of managers is kind of what matters. And I've learned how to start managing managers much better. I've gotten good at it.
All right. Last time we spoke, I have a number of follow-ups here. Let me ask the easy one, might not be the easy one to answer, but the easy simple question first, which is why a daily metric as opposed to a different time interval like a weekly, for instance, in terms of revenue or something else.
There's two reasons for daily stuff. One is we're really big believers that you should be using Duolingo every day. You don't have to use Duolingo every day to learn a language, but we really want to build a habit. And the best habits are daily, I think, because it's kind of hard to have these habits that are like, 'Well, every 3 days I do something.' Like it's just kind of hard. It's much better to have a habit like brush my teeth, just do it every day. So, we try to build habits every day. And so, for us, this is one of the reasons why we look daily. The other reason why we look daily is that we run a lot of A/B tests. And looking at the daily time frame allows us to move faster because imagine that our time frame was a month. We said, 'Look how many people do X in a month.' We have to wait a whole month to figure out what that number is.
Right. So and that's—is that because you're just controlling for confounding factors if you're running, if you're changing multiple variables at the same time? It just becomes too difficult to attribute causation. So the lag time just becomes long.
That's exactly right. So whereas if you look at daily stuff, it's much faster. So for example, for revenue, we look at daily bookings, to be precise, which is basically the amount of money that comes into Duolingo. We look at that a lot and we optimize that, and it really is nice because you can see, for example, if we make a change, you can see tomorrow what happened in the change today. You can already see, 'Ah yes, daily bookings increased by, you know, 10%.' You can see it in one day. Those are the two reasons why we look at the daily. I mean, we've talked about starting to do hourly. We haven't quite done that. I know one of the difficulties with hourly is there's a pretty big difference in terms of times. I mean, for example, 9:00 p.m., around 9:00 p.m. is the most popular time to learn a language, but there are times, for example, 4:00 a.m., there's just not very many users. And so, looking at hourly makes it a little harder.
So, before we move on, I'd love to just clarify one thing that you said, and that is related to the micromanaging and whether that point is at 10 people, 20, 30, but it was related to being pre-product-market fit. Would it be safe to say that if you have product-market fit, let's just say you've created a business model and a business that is generating revenue, that has found its match, that it would potentially make sense to develop people sooner than the 10 people, 20 people, or whatever the headcount might be.
I think that makes sense. It just, you know, hopefully you're not in a period where you're doing this kind of zero to 20 employees for eight years or something. Hopefully that period only lasts one or two years, and in that time you probably just need to be trying to figure out product-market fit if you're going to be in that for several years. I do think it starts making sense developing people.
To, let's say, just hop back to our first conversation. There were two books that you mentioned as being particularly helpful, one or that you would recommend maybe both. One was 'Zero to One' by Peter Thiel and the other was 'The Hard Thing About Hard Things,' if I'm getting the title right, I can never remember this title, by Ben Horowitz. And I'm wondering if now, flashing forward from 2016 to present, you've gone from 60 to 600 or so people, a lot has changed, you've gone from private to public, are there any other books or resources or thinkers who you have found particularly helpful or helpful at all in the last handful of years.
One book that I really liked, and we make all managers read it at Duolingo, is—it's an oldie. It's an oldie but a really good one—is 'High Output Management' by Andy Grove.
Yeah. It's just like a how-to manual. Like he doesn't bother with like his life story or whatever. I mean there's a little bit of that, but it's really, it really is just very practical. It's like, look, I have found that when you're doing performance reviews, you can talk to the people before giving them the review or after. You should always just do it after. Like it's just stuff like that. Like it's just these little things that are just like, just I have found that this is better than that. And there's just a lot of, you know, here another one that I—it's just kind of off the top of my head—is like when somebody tells you that they want to quit, drop everything you're doing. If you don't want them to quit, if you want them to quit, yeah, let them quit. But otherwise drop everything you're doing and make this your highest priority because one of the main reasons people quit is because they don't think that you think they're important. And so it's just very, very valuable advice like that that is just super applicable immediately to everything you're doing. So I really think that book has helped me a lot to become a better manager.
We are going to discuss a number of facets of Duolingo and new developments and so on. Before we get to that, I asked a number of questions last time that I think you may have new answers to. So, I'm going to just ask those. I think there are four at least in front of me, and then we'll zig and zag back into Duolingo-related things. So the first is what is one of the best or most worthwhile investments you've ever made? That could be an investment of money, time, energy. Pick your currency.
The easy answer for that is Duolingo, although, I mean, I've been at this for so long that it was not an easy investment. Obviously, has been very good. Um, you know, I'll give a counterintuitive one that has really helped me. I'll say it's a weird one. I have an aging mother, and I think a lot of people my age have aging parents. I used to worry a lot about her because she used to live in Guatemala. And you know, Guatemala is a dangerous place, and I'm like, uh, she's pretty old. I mean she had me at an old age. She had me at age 42, so she's 85. And she used to drive still in Guatemala. And it was like this whole thing. And I worried a lot. A lot of my mental energy was—it's not 80% of my mental energy, but it was a significant fraction of my mental energy was just worrying that she was okay. And the investment that I did is she moved in with me. And it's a weird thing. I mean, my house is large enough that this is fine. We get along well and everything. But I feel so much better about it. I mean, I'm just like, now I'm not worried about her. She's there. She's thriving, and she essentially keeps to herself. It's not that like she bothers me all that much, but I just—I'm just not worried about it. This has made me well—I freed up, I don't know, call it 4% of my mental energy that I was spending on that is freed up. I just have more time to worry about other things.
The next question is, in the last 5 years, what new belief, behavior, habit has most improved your life? And at the time, the first conversation we had on the podcast, it was starting Duolingo in Pittsburgh. So, kind of a slightly oblique answer to the question, but I'm wondering if you have a different answer to that or an additional answer to that now.
I think a personal one, if we go on the personal one, a habit that has really changed my life in the last few years is strength training. I used to always kind of just do cardio to work out and that was the main thing. But one thing that happened is I was—I don't know the year, maybe 2017 in January. First of all, uh, I don't love the winter in the northeastern United States. I am from Guatemala. I'm not used to this. This makes no sense. How could you have a place where people just die outside if they don't do anything? It's kind of ridiculous. But anyways, I don't love the winter. It was January. I was not having a great January. I also got sick. I had some sort of like a flu, and I was not doing super well. And I went to the office finally after all this. And I guess I looked—I didn't look so good, I guess. And somebody that worked at the office who's worked with us for a very long time, Dennis, looked at me and he's like, 'Luis, you need Gary.' And I'm like, 'What? I don't know what you're talking about.' He's like, 'What are you doing tomorrow at 6:30 a.m.?' I'm like, 'Nothing.' Gary will show up. Anyways, turns out Gary shows up. Gary is his personal trainer, Dennis's personal trainer. Gary is kind of quite a personality. He used to be the strength trainer for the Pittsburgh Steelers and then the strength trainer for University of Pittsburgh basketball team. And it's just kind of—and now he just has this roster of clients of like all these rich people from Pittsburgh who are not athletes. Certainly I was not an athlete. And Gary shows up and he's like, 'All right, I'm going to make you stronger.' And you know, he gives you some weights and then he laughs at how weak you are. He like laughs. I'm like, 'What are you supposed to make me feel good?' He's like, 'Oh my god, you can't even do like whatever, 40 lb dumbbells for bench press. What are you doing?' And so he laughs at you, and I've been with him since like 2017. It has really changed my life. I used to have all kinds of aches and pains. No longer. It turns out, you know, if you know a lot more about this than me, but I think if you start working out, doing strength training after age, call it 35 or so, it's pretty hard to build big muscle. So, it's not like I've built humongous muscle, but I'm just a lot healthier and certainly a lot stronger. And so, this has really changed my life. So, Gary's awesome. Big, big fan.
Wow. I guess Dennis deserves some significant thanks, too, for just taking the initiative.
Oh, Dennis is also awesome. No, he really—he just looked at me and I must have looked like hell. He just looked at me. He's like, 'You need Gary.'
Enough said. Not going to elaborate. Yeah. How many times per week do you do resistance training now?
Three times a week. We'll give this a shot. We'll see if this question goes anywhere, and then we're going to kind of zoom back to Duolingo and related topics. What purchase of $100 or less has most positively impacted your life in recent memory? And of course, listeners love specifics.
Easy. An electric toothbrush.
Ah, good. I got the Oral-B. This changes your life. I mean, it just—you can't go back. You can go back. And I'll tell you, when I started dating my current significant other, one of the first few things that I did was I saw that she was kind of slumming it with like a regular toothbrush. So on Amazon, ordered her, my first gift to her was an electric toothbrush. She said to this day—we've been together for longer—I've given her significantly more expensive gifts. To this day, best gift I've ever given her.
It really just changes your life. If you're not brushing with an electric toothbrush, you're doing it wrong.
You know, I didn't expect this to come up in this conversation, but I'm going to plus one that and mention that I recently went to a new dentist. Very highly recommended by a separate doctor I work with. Apparently, top of the top as far as dentists go. I was nervous to go because I'd not been to a dentist since COVID. Now, this is probably flashing back.
Yeah. Common. Common. So I was terrified that it was just going to be a complete, you know, warfield of issues. But it was mostly fine. But the point he made to me was with his recommendations, he said, 'Number one, you need to use an electric toothbrush of some type because it will prevent you from applying too much pressure.' And there are other benefits to it, but he—I apparently brush my teeth like I'm trying to take old grout off of tile or something and I just destroy my gums. And it's not that I'm trying to push really strongly, but I do nonetheless and end up with bleeding gums and all of this nonsense. And since using an electric toothbrush, as far as that problem goes, problem solved.
It's one of those where you like just don't realize because people had told me this before, but you just don't realize how much it changes your life. Another similar one. It's not a $100, but it is really life-changing. Is a Japanese toilet.
Okay. Did you get the full Japanese toilet or the... Okay, I got the full total Japanese toilet. This again, more than 100 bucks, but life-changing. It is really—now it does cause a problem which is I now really—it's not like I can't—I really don't like going to the bathroom anywhere but my house because it's just a significantly better experience.
All right, for people who may be curious, I have a couple of friends who are obsessed with Toto toilets. Do you have a specific model? Can you recall which you decided on?
Toto Neo Rest and it is—hold on—it is a specific one, the Neorest NX1.
It is, let me tell you, let me—obviously the main feature is that it uses kind of like the bidet feature where it cleans you with water, which after you start thinking about it, it is barbaric that people in the United States and most Western countries do not use water to clean themselves. It really is after you think about it.
Well, I like the adjective. I'm a fan.
Well, just think about this. I mean, if you were to touch with your finger some cow poop.
Would you not wash your hands with water? Like, would you just use a towel to like... No, you wouldn't. That just... because you would feel like it's still dirty. And so, think about that. So, it obviously has that very feature, but it has all kinds of other stuff, including like white noise so that other people can't hear you. Like, it just has features. It is good. All right. So, now let me give an alternative here because I mean, I'm guessing the Neo, this particular model, is—I mean—it's probably at least a few thousand.
Yeah, it's probably at least a few thousand.
I will give people an alternative if they want somewhere between the Bugatti of toilets and the barbaric practice that you already outlined. You can get, and actually I was given as a gift, a Washlet Toto seat. So it effectively replaces your toilet seat on your pre-existing toilet, and it's a few hundred. You plug it into a wall, and it has a lot of the same features. So I'll just mention yes,
it is pleasant to use. And you know, I never thought I would have a Japanese toilet, or I suppose a hybrid Japanese toilet, which I have now. But when I was in Japan, when I was 15, there's some basic features. I mean, when it is freezing, and it gets cold in certain parts of Japan, when a seat warmer—
Yeah. When it's freezing and you go in and you sit down and it's warm,
that has a material impact on your day-to-day experience.
