Duolingo
If you haven’t already, join 10,080 other business lovers and subscribe for free:
I use Koyfin every day. From staying on top of company news to the most advanced screener you can find, it’s an incredibly good investing tool. My personal favorite is the customized financial templates. Using Koyfin’s shortcuts, I can pull up revenue, gross profit, EBIT, FCF, net debt, goodwill, ROIC, and working capital dynamics for any company in seconds.
I also have a huge watchlist that helps me with earnings dates and allows me to easily find new transcripts and filings. Rather than having one charting software, one for filings, one for transcripts, and one for financials, Koyfin has it all. And the best part is you can customize it exactly to your preferences. I rarely am passionate about investing products but this is one I strongly believe in. Because you’re a Business Breakdowns reader, use the button below to get 10% off on a plan today!
Duolingo was started in 2011 by Luis von Ahn and Severin Hacker (an amazing name for a software dev). Luis and Severin met at Carnegie Mellon, where Severin was a standout PhD student for professor von Ahn. In fact, Luis has quite an interesting background. In 2007, he started a company called reCAPTCHA and sold it to Google. Yes, he is responsible for the fact that you can’t enter a website until you differentiate between a bus and car approximately six times, mess it up, and start over again. The original premise for CAPTCHA actually solved two problems. One, it limited bots from overwhelming websites. And two – you probably remember this when it was crazy letters before it was pictures – these letters were directly from old books that couldn’t be optically recognized. So CAPTCHA started as a crowdsourced way to make accurate digital books through improved optical recognition. What’s really interesting is that this sort of led to the idea of Duolingo. After he sold CAPTCHA and became a professor, Von Ahn recruited Hacker to help him find a way to translate websites at scale. The duo (I promise I won’t make a bad pun) decided to use the same concept from CAPTCHA to crowdsource accurate translations. So they made a web app that would help people learn English while also providing solid translations. Then, they sold these translations to publishers that were looking for a cheap way to translate their web pages into different languages. The fabric of how CAPTCHA and Duolingo are intertwined is fascinating as they share the same foundation of crowdsourcing.
However, as translations improved over several years, Luis and Severin changed business models to a freemium app where users could pay to stop seeing ads. This enabled fast growth as the app was free and there was an amazing amount of content. This also aligned with the company’s mission which is to develop the best education in the world and make it universally available. Notice that it doesn’t say anything about language learning. Duolingo, despite a name that alludes to language learning, is focused much more broadly, long term, on general education. More recently, the company announced a large push into math and music over the next decade. These categories make a lot of sense as they are global languages of a different kind.
Nowadays, the company makes 75% of its revenue from subscriptions to its language learning app. The rest is split between advertisements to free users and virtual items that users can buy to get perks in the app. The company also has small revenue like from an English test that can be paid for to prove English competency for college exams, etc.
If you’ve ever used the app, it’s a well-oiled machine. It uses social media techniques like streaks to retain users at very high levels for an app, let alone an app that is making the user think. Many users, including my wife, will finish their daily lesson for the sole reason to “maintain their streak.”
Von Ahn and team are maniacally focused on decreasing customer acquisition and increasing retention. Through sophisticated data analysis, every marketing channel and app function is tracked to improve both of these metrics. And thus far, these efforts are paying off. The company has grown monthly active users from 24 million to over 85 million over the past 5 years. Meanwhile, the proportion of daily active users has improved from 23% to 30% and the percentage of premium subs to MAUs has gone from 3% to over 7%. Management thinks they can reach mid-teens on this metric, which would be more in-line with a dating app like Tinder. Duolingo has had 500 million users make an account since inception, which is a staggeringly large percentage of the global population that has a smartphone.
Personally, I was a bit skeptical that the company would be able to maintain its growth rate after COVID as revenue grew 130% in 2020 but Duolingo has grown revenue over 40% for each of the last three years. The company just keeps on executing on the back of really strong paid subscriber growth.
The company is also fairly profitable now too despite gross margins that will never go much beyond 75% since app stores take more than 20% of in-app purchases. Free cash flow is positive even after stock-based comp which can’t be said for even some of the best SaaS companies. Free cash flow margins pre-SBC were 24% for 2023 as the company improved its mix of annual payments instead of monthly subscriptions. Another very important thing to point out is that the company spends more on R&D than S&M. In 2023, Duolingo spent nearly $200 million on research and development compared to just over $75 million on sales and marketing. That is not even close! This is firmly ingrained into the ethos of the company – it’s truly product led. It is working on all sorts of things but primarily this is spent on people and improving the quality of the content. Duolingo quickly started using GPT-4 to improve how its animated characters interact with users, making the whole experience more realistic. I think the company will eventually come out with a personalized AI tutor as well that users can speak to. This would unlock major improvements in language learning and Duolingo is well positioned to benefit from these trends, even though there are also worries that language learning won't be necessary because of AI translation. It’s an interesting point but I think it misses a lot of why people learn another language. It’s very personal. Even right now, you could use voice translation to speak with someone, but it won’t be a very relational experience. It sort of reminds me of Zoom – sure, it’s more convenient but there is something about learning a language that honors the person you’re speaking with. I also think people just enjoy learning something when it is fun.
Pretty soon here, the company will reach GAAP profitability and the balance sheet is very strong with nearly $700 million in net cash. With a fantastic, visionary founder at the helm, a large market, and amazing metrics all around, the real question is valuation. But that’s not why you read Business Breakdowns. We do that in our other services :)
Just kidding, I don’t want to cop out that hard. The enterprise value is about $9 billion, which probably implies that Duolingo has some more success outside of language learning. Will there really be a market of 200 million MAUs? Maybe? At $150 ARPU, that’s a serviceable market of $3 billion in revenue or 6x the current size. I truly don’t even really know what the upper limit for paid subs is. Right now, you may be surprised to learn that the company is on the same paid sub trajectory as Netflix, only 8 years later. From 2011 to 2016, Netflix went from 24 million global subs to 89 million. That’s really close to Duolingo actually. This next phase, however, will likely be much more difficult. It’s a whole lot easier to sit on your butt and watch Netflix than it is to learn a language everyday. Netflix has about 250 million global subs so I don’t think Duolingo will get close to eclipsing that over the next decade. At $3 billion in revenue and 30% FCF margins (company’s long-term target) and a 20x multiple, that’s just over a double from today’s prices. The questions then are, how long will that take and does $150 ARPU imply too much pricing power? Since there is a lot less competition than streaming, maybe there is some leverage but the nature of the products are very different. You could choose any app-based comp but I think Netflix is interesting because Duolingo really is more of an entertainment app since it’s so gamified. I’ll let you figure out the answers to those difficult questions but I think Duolingo is a special company with a special founder running the show.
If you enjoyed this, please turn that ♡ into a ♥️. This will help others find this newsletter. Thank you so much 😁
How Can We Help?
Looking to become a partner in our long-only strategy?
Interested in advertising to more than 10,000 investors?