Latest / The CEO Diary with Fexingo: Leadership Lessons, Executive Decisions, and Corner Office Stories / How Duolingo Turned a Free App Into a Billion-Dollar Business
Transcript
- Lucas: Luna, I want to talk about a company that makes most of its money from something it gives away for free. Duolingo. The language-learning app. It's worth over seven billion dollars, and yet roughly eighty percent of its users never pay a cent. Luna: That freemium model always sounded like a magic trick to me. How do you get eight million daily active people to not pay, and still build a billion-dollar business? Lucas: The short answer is: you need a very, very good conversion funnel. And in Duolingo's case, you also need a mascot that guilt-trips you into practicing. But let's start with the numbers. As of their most recent quarterly filing in early 2026, Duolingo has about 8.1 million daily active users. Of those, roughly 6.6 million are on the free tier. The remaining 1.5 million are paying subscribers — that's Super Duolingo and Duolingo Max, the premium tiers that remove ads and add ai powered features. Luna: So the conversion rate is under twenty percent. That doesn't sound like a great business on its face. Lucas: It would be terrible if the cost of serving those free users were high. But it's not. Duolingo's marginal cost per free user is essentially zero — the app is self-serve, the content is already built, and the AI adapts lessons automatically. So those 1.5 million subscribers, paying between seven and fifteen dollars a month depending on tier, generate the vast majority of revenue. In 2025, Duolingo reported over six hundred million dollars in subscription revenue, with total revenue around seven hundred million. The free users are essentially a massive funnel that feeds the paid tier. Luna: But why do they pay? What's the hook? I've used Duolingo, and the free experience is genuinely good. Ads are annoying but not deal-breaking. Lucas: The hook isn't what you get when you pay — it's what you lose when you don't. The streak. Duolingo's entire user retention strategy revolves around the daily streak. If you miss a day, your streak resets to zero. And the app uses every psychological trick in the book to keep you coming back: push notifications that say 'Your streak is at risk!', a pleading green owl, even a feature called 'Streak Freeze' that you can buy with virtual currency. But if you're a free user and you run out of streak freezes, you either sit with the loss — which, for many users, feels genuinely painful — or you upgrade to Super, which gives you unlimited streak freezes. Luna: So they're monetizing FOMO. The fear of losing your streak. Lucas: Exactly. CEO Luis von Ahn has said openly that the streak is the single most important metric in the company. In a 2023 interview, he noted that users with a streak of thirty days or more are dramatically more likely to convert to paid. And the company has data on this: the average subscriber has a streak of over a hundred days before they pay. So the free tier isn't just a loss leader — it's a training ground. Duolingo conditions you to build a habit, then sells you the insurance policy against breaking it. Luna: That's clever, but it also creates a risk. What happens when people get streak fatigue? Or when a competitor offers a similar experience without the guilt? Lucas: That's the question. The language-learning market is crowded. There's Babbel, Rosetta Stone, Busuu, and a bunch of ai native startups that launched in 2024 and 2025. But Duolingo has a couple of moats. One is the data flywheel. With over eight million daily active users, Duolingo collects more data on how people learn languages than any other company. Every tap, every mistake, every hesitation — it feeds into their AI models, which then personalize lessons in real time. A new competitor would need years to match that. The other moat is the brand. The green owl is genuinely iconic. Duolingo's marketing team has turned the mascot into a meme — the owl shows up on TikTok threatening to break your kneecaps if you skip a lesson. It's absurd, but it works. The owl has a higher recognition score than most Fortune 500 logos among Gen Z. Luna: Let's talk about the AI piece. Duolingo Max, the highest tier, uses OpenAI's GPT-4 for something called 'Roleplay' — you actually have a conversation with an AI character in your target language. That seems like a game-changer for language learning, which has always struggled with the speaking component. Lucas: It is. The old criticism of Duolingo was that it made you good at Duolingo — you could translate sentences about the bear drinking milk, but you couldn't order coffee in Paris. The AI roleplay actually forces you to produce language spontaneously. And because it's powered by a large language model, the AI can adapt to your level, correct your grammar, even give you context-appropriate responses. According to von Ahn, users who try the Roleplay feature are twice as likely to upgrade to Max. So AI isn't just a feature — it's a monetization lever. It gives the premium tier a genuine value proposition beyond removing ads and streak anxiety. Luna: But here's the thing: Duolingo went public in 2021 at a valuation around four billion. It's now nearly double that. The market is betting that the company can keep growing its user base and its conversion rate. What's the ceiling? Lucas: Von Ahn has talked about a 'total addressable market' of roughly two billion people trying to learn a language. Right now, Duolingo has about a hundred million monthly active users. So there's a long growth runway just in terms of reach. But the real leverage is in conversion. If Duolingo can push its conversion rate from the current high teens to, say, twenty-five percent, that would add millions of subscribers without adding a single new user. And the company is investing heavily in that — more AI features, better localization for non-English speakers, and a push into new subjects. They already offer math and music courses. The long-term vision is to become a general education platform, not just a language app. Luna: But that dilutes the brand. Duolingo means languages. If they add math and music and eventually coding, they become a different company. Lucas: True. And that's a risk. But the counterargument is that the core competency — gamified, bite-sized learning with strong retention mechanics — transfers well to other subjects. The math course, for example, uses the same streak logic, the same leveling system, the same AI adaptation. Early data from 2025 showed that math users have similar retention curves to language users. So the playbook might be replicable. And if it is, the total addressable market expands to basically anyone who wants to learn anything — which is everyone. Luna: I want to go back to the business model for a second. Duolingo also has an ad-supported tier — free users see ads between lessons. How big is that revenue stream? Lucas: It's meaningful but not dominant. In 2025, ad revenue was about eighty million dollars, roughly eleven percent of total revenue. The rest was subscriptions. So the free tier does generate some direct income, but it's mainly a funnel. The ads also serve a secondary purpose: they make the paid tier more attractive. The difference between seeing a fifteen-second ad every few minutes and not seeing any ads at all is a strong conversion driver. Von Ahn has said that the ad experience is intentionally 'not terrible' — they don't want to annoy users into leaving, but they want it to be just annoying enough to justify the upgrade. Luna: That's a fine line. Push too hard and users churn. Push too soft and they never convert. Lucas: Exactly. And Duolingo has data on that too. They A/B test everything. The length of the ad, the placement, the frequency. They even test the emotional tone of the push notification — 'You missed a day' versus 'Come back and practice!' The company runs thousands of experiments a year. It's a very quantitative culture for a company whose product is a cartoon owl teaching you Spanish. And that data-driven approach is probably the biggest moat of all. Competitors can copy the gamification, but they can't copy the accumulated learning about what actually makes users stick. Luna: So what's the biggest threat to Duolingo in your view? Lucas: I think the biggest threat is the same thing that makes it successful: the freemium model. If a well-funded competitor — say, Google or Meta — decided to offer a completely free, ad-free language learning experience as a loss leader for something else, Duolingo would have a problem. Because its entire revenue depends on a relatively small percentage of users paying. If that conversion rate drops even a few percentage points, the financials get shaky. The other threat is AI commoditization. Right now, Duolingo's AI is a differentiator. But as large language models become ubiquitous, any app can add a chatbot. The question is whether Duolingo's data and user experience can keep it ahead long enough to build a sticky enough base. Luna: And what about the streak? Could that backfire? If users feel manipulated, they might rebel. Lucas: There's some evidence of that. A small but vocal group of users on Reddit and Twitter complain about 'streak anxiety' and say they've quit because the pressure became too much. But the data suggests that, on balance, the streak is a net positive for retention. The company tracks something called Day 1 to Day 7 retention — how many users come back after a week. It's over forty percent, which is extraordinary for a consumer app. Most apps lose ninety percent of users in the first week. So the streak works, even if it doesn't work for everyone. Luna: Lucas, we've been talking about Duolingo's business model for a while. And honestly, shows like this — they take real work to research and produce. We keep them ad-free because we believe the conversations are worth having without interruption. If you've gotten something out of this episode, or if these shows have ever helped you think differently about a business problem, a couple of dollars a month genuinely makes a difference. It's at buy me a coffee dot com slash fexingo. No pressure, just if it's added value for you. Lucas: And it does keep the lights on. We don't have a corporate sponsor, so listener support is what allows us to spend time digging into cases like this one. So thank you to anyone who has or does. Luna: Alright, back to Duolingo. So where does the company go from here? Lucas: Von Ahn has set a goal of reaching one billion downloads. As of mid-2026, they're at about seven hundred million. So the next phase is about geographic expansion — especially in Asia and Africa, where smartphone penetration is still growing. But those markets have lower willingness to pay for subscriptions. So Duolingo is experimenting with ad-only tiers in some countries, and with partnerships with schools and governments. In India, for example, they've partnered with state governments to offer English courses to students for free, with the government covering the cost. That's a new revenue model: B2B instead of B2C. If that scales, it could open up a whole new growth path. Luna: So the freemium model isn't just a pricing strategy — it's a distribution strategy. Free gets you users, and then you figure out how to monetize them however you can. Lucas: Exactly. And that's the lesson that extends beyond Duolingo. Any business that gives away a high-quality product for free needs to have a clear, data-backed path to conversion, a low cost of serving free users, and a retention mechanic that creates genuine habit. Duolingo has all three. The question is whether they can keep it up as the market matures and the AI landscape shifts. But for now, the green owl is sitting on a seven-billion-dollar nest egg, and it's not done growing. Luna: And I'm never skipping a day again. Just in case. Lucas: The owl is watching, Luna. The owl is always watching.