Latest / Startup Stories with Fexingo: Conversations About Founders, Funding, and Building Companies from Zero / How Duolingo Gamified Language Learning Into a Public Company
Transcript
- Lucas: Duolingo is a language learning app with a cartoon green owl that nags you to practice your Spanish. It is also, improbably, a $10 billion public company with over 100 million monthly active users. Luna: Improbably is the right word. I mean, language learning was never supposed to be a huge business. Rosetta Stone tried, went public, struggled, got taken private. Lucas: Right, exactly. Rosetta Stone peaked at around 4 million subscribers. Duolingo has more than 10 times that many monthly actives, and it's still growing north of 40 percent year-over-year. The question is: how? Luna: Well, the obvious answer is gamification. But that's become a buzzword. I want to know what Duolingo actually does differently. Lucas: So let's start with the founder. Luis von Ahn is a Carnegie Mellon computer science professor. Before Duolingo, he co-created CAPTCHA and reCAPTCHA — those distorted text tests that prove you're human. He sold reCAPTCHA to Google in 2009. So he understands user psychology at a deep level. Luna: He basically trained millions of people to digitize books for free, and they thought they were just proving they weren't bots. Lucas: Exactly. He's good at getting people to do things they wouldn't otherwise do. With Duolingo, the insight was: language learning is hard and people quit. So don't fight that — design the product to make quitting feel like a personal failure. The daily streak is the key mechanic. If you miss a day, your streak resets. People have 2,000-day streaks. That's almost six years. Luna: And they get notifications. The owl sends things like 'These reminders are not working. You leave me no choice but to send Duo the Owl to your house.' Lucas: It's aggressive, and it's funny. But it works. The company has publicly said that notifications are one of their highest-leverage growth channels. They've A/B tested everything — the time of day, the tone of voice, the emoji. They found that guilt-tripping actually outperforms encouragement for re-engagement. Luna: That's dark. But also, I get it. I've personally been shamed into conjugating verbs at 11 p.m. Lucas: So the product is sticky. But the business model is also interesting. Duolingo is freemium with ads. The majority of users never pay a cent. The company makes money from ads and from Duolingo Super, the subscription tier that removes ads and adds features. About 5 to 6 percent of monthly active users convert to paid. Luna: That conversion rate sounds low, but for a free product with a massive user base, it's actually healthy. Especially because the subscription is relatively cheap — about $13 a month in the U.S., often discounted. Lucas: Right. And they've been adding revenue streams. Duolingo English Test is a proctored online English proficiency exam accepted by thousands of universities. It costs $59, compared to TOEFL at over $200. That's a growing business. They also have Duolingo Math and a literacy app for children. Luna: So the core app drives the flywheel, but the adjacent products monetize more directly. That's smart. Lucas: Let's talk numbers. In 2025, Duolingo reported around $750 million in revenue, with a compound annual growth rate above 40 percent since 2020. They turned profitable on a net income basis in 2024 and have stayed profitable since. Gross margins are around 75 percent. Luna: Those are SaaS-level metrics. For a consumer app that gives away most of its product for free. Lucas: Yeah. And the market cap reflects that — it's around $10 billion. That's a multiple of roughly 13 times trailing revenue, which is not cheap but not outrageous for a company growing that fast with that margin profile. Luna: If today's episode was useful to you and you want to keep it ad-free, buy me a coffee dot com slash fexingo helps. Lucas: Yeah, listener support really does make this possible. And we keep it simple — no ads, no sponsors. Buy me a coffee dot com slash fexingo. Anyway, back to Duolingo. Lucas: So one thing I find fascinating is how they kept the app lightweight. For years, Duolingo didn't have a dedicated desktop team. The whole company is mobile-first. The app is less than 100 megabytes. They deliberately avoid feature bloat. Luna: That's rare in edtech. Most learning platforms try to add video, live classes, community forums, and then they become a mess. Duolingo is laser-focused on that 10-minute lesson. Lucas: And they use AI heavily. The company was early to adopt GPT models for generating lesson content and for conversation practice. They have a feature called Duolingo Max that uses OpenAI to simulate real conversations. That's a premium upsell. Luna: So AI is both a product feature and a cost-saving tool. They generate exercises automatically, which keeps content production scalable. That's a huge advantage over traditional publishers. Lucas: The risk, of course, is that a competitor builds a better AI tutor. There are already startups doing one-on-one AI tutoring in math and science. But Duolingo has the data and the habit loop. That's hard to replicate. Luna: And the brand. That owl has become a meme. It's cultural shorthand for 'you should be practicing your French.' Lucas: Right. So the moat is partly behavioral — the streak, the notifications, the sunk cost of 500 days — and partly brand. But it's also network effects. The more people use Duolingo, the more data they have to improve the AI, the better the lessons get, the more people stay. Luna: It's a classic data network effect. Every interaction feeds the model. Lucas: Exactly. And that's a defensible loop. So when we talk about building a company from zero, Duolingo is a great case study in starting from a genuinely novel insight — not 'let's make a better textbook' but 'let's make language learning feel like a game, and then figure out the business later.' Luna: And the business came. It just took a while. They didn't even start running ads until 2015, years after launch. Lucas: Right. The early focus was entirely on growth and engagement. Monetization was an afterthought. That's the opposite of most startups today, which try to charge from day one. Luna: So the lesson might be: if you build something people genuinely want to use every day, you can figure out the revenue later. Lucas: I think that's part of it. But you also need a founder who can afford to wait. Von Ahn had Google money. He wasn't desperate. He could optimize for user love over revenue. That's a luxury most founders don't have. Luna: Fair point. So maybe the real lesson is: if you can, build a product so good that people build a habit around it. Then the business model reveals itself. Lucas: And if you can't, at least make the owl really, really persistent. Luna: I think I have a notification from Duo right now, actually. Lucas: You probably do. That's the whole point.