Duolingo ($DUOL) Deep Dive
Does the green owl have a future?
Welcome to Leeder Capital! I have previously covered, Evolution AB, Airbnb, Crocs, Paycom, Xpel, Moncler, Charter Communications, and Judges Scientific.
Subscribe below to not miss out on future deep dives!
Disclosure: I am long shares of Duolingo.
(Note: this article should not be considered investment advice; please see the full disclaimer at the bottom.
Short History
Luis Von Ahn grew up in Guatemala, raised by his mother and grandmother during a period of turmoil. He was fortunate to receive a good education, while the rest of the kids he grew up with didn’t; he saw the divide in the opportunities available to them.
“She spent her entire wealth, really, her entire salary and everything on my education, so she basically gave me a rich person’s education, even though we were not rich.” Luis Von Ahn
By the age of 8, he had taught himself to code, and by 13, he knew he wanted to be a professor of computer science. He moved to the US to study Maths at Duke, and at 28, he became a professor at Carnegie Mellon University. Just a week later, he was awarded the MacArthur Genius Grant for his work in human computation (2006). From his time as a Grad student onward, Von Ahn built tools on the side, using games to solve problems that computers could not. This data could then be used to teach computers how to solve the same problems. In 2006, he sold his first venture, the ESP game, to Google, which helped lay the foundation of Google Image Search.
In the early 2010s, he spent a short stint at Google after selling his second company, RE-CAPTCHA, in 2009. After a year and a half, he left Google to start Duolingo.
Duolingo initially started as an academic project between him and his Phd student, Severin Hacker, to provide equal access to education for everybody, but due to the slow process of getting a grant, it turned into a company: Duolingo was founded in 2011 with the mission “to develop the best education in the world and make it universally available”, inspired by Von Ahn’s firsthand experience growing up in Guatemala.
When deciding on the initial subject, Von Ahn pushed for languages because learning English had unlocked so many opportunities for him.
After a year and a half of development, Duolingo launched in 2012.
“From the outset, Duolingo’s approach set it apart. Rather than traditional classroom-style instruction, it was structured like a game. Users would progress through bite-sized lessons, earn points, and level up. This was quite different from the dry educational programs from CD-ROM software of the past. With a colorful and easily navigated UI and a clear sense of progression, it was much more akin to a game than a lesson. But beneath the friendly graphics, serious thought and data analysis went into pedagogy and psychology to keep users engaged while learning.” - Quatr
The app immediately found product-market fit; downloads and usage grew rapidly: by the end of 2013, Duolingo had 10 million users, and Apple had just named it iPhone app of the year.
Over the following years, the company focused solely on improving the app: developing new courses, adding gamification features (streaks, social features, etc.), and optimising engagement, before eventually monetising it in 2017, 5 years after launch. This intense focus on improving the product before monetisation led to rapid user growth, and by 2021, Duolingo had been downloaded over half a billion times!
While high growth rates were inevitable in the early years, Duolingo has sustained impressive growth since going public (2021) while becoming consistently profitable. Users have grown to 135 million monthly active users (MAUs, 32% CAGR) and 50 million daily active users (DAUs, 48% CAGR).
Overview
Today, Duolingo is the global leader in mobile language learning. Its flagship app, Duolingo, is the top-grossing app in the education category on both the Apple App Store and the Google Play Store; management believes it has 80% market share of all online MAUs.
From just four languages at launch, the company has expanded significantly, offering over 100 language courses (most of which teach English from a given language); today, the app teaches 42 languages, of which the top 9 account for the vast majority of demand. 50% of users learn English, 20% Spanish, and 10% French. German, Italian, Japanese, Korean, Portuguese, and Mandarin make up the remaining languages in the top 9. Over time, the company has also expanded beyond language learning, first to Duolingo ABC (a separate early-development app) and then into Math, Music, and Chess. Subject expansion has been successful; however, the vast majority of traffic is still concentrated on language learning.
Users
Breaking down the 135m MAUs and 50m DAUs, users can be split into 2 relatively equal segments:
Hobbyists: People who enjoy learning a language/ subject.
People who need to learn a language (mainly English) to access future opportunities, such as education and higher-paying jobs.
The user base is broad across geographies and demographics, similar to the industry as a whole. As previously mentioned, the largest concentration is among English learners; however, this penetration is significantly below the industry average of 80%, likely due to the apps’ concentration of beginners.
Revenue streams
The US accounts for just under 40% of revenue; beyond that, Duolingo does not disclose geographic breakdown. However, piecing together several comments, it appears that the majority of the remaining 60% comes from Western Europe, Japan, and Korea.
When most companies face the challenge of making money, greed kicks in: they raise prices, increase ad load, and make the free product less attractive to incentivise paid conversion. For the most part, this has not happened at Duolingo; the free product has remained just as good, consistent with its mission of making education universally available. This acts as a significant headwind to paid conversion, particularly in geographies such as Asia and South America, where users will not pay unless they have to. As a result, Duolingo is leaving money on the table to prioritise the user experience.
In a decision between increased monetisation and user growth, Duolingo chooses the latter.
Duolingo makes money from two products: The Duolingo app (95%) and the Duolingo English Test (5%). The flagship app is monetised through a freemium model, generating three revenue streams: Subscriptions (84%), advertising (8%), and in-app purchases (4%).
Over time, the company has added additional subscription tiers and rolled out regional pricing (lower pricing). The current subscription tiers look like this:
Free tier: Over 90% of users are on the free plan; they have access to all courses and the vast majority of app features, with some added friction: mainly ads; a single ad is displayed at the end of each lesson. The energy system is another source of friction, lessons use energy, if users run out of energy, they can view additional ads to replenish it. Despite accounting for the vast majority of users, Advertising revenue accounts for just under 8% of the business. Free users are also the largest source of in-app purchases.
Super Duolingo: This is the original paid tier, costing (in the US) $12.99/month or $84/year. This tier removes the friction from the free tier: removes ads and unlimited energy. Super Duolingo can be subscribed to individually or as a family plan, allowing up to 6 users on the same plan. Family plans have significantly higher retention.
Duolingo Max: rolled out in 2024, Max offers all of the Super features plus AI features such as ‘Explain my answer’ and ‘Roleplay’. This tier is priced at $30/month or $168/year.
On the whole, people willing to subscribe tend to have higher incomes, while those unwilling to pay tend to have lower incomes. As a result, paying subscribers help make a better app, which provides better education to those who can’t afford it.
Growth drivers
Growth has been driven by three factors: the growing number of users, an increasing share of users willing to pay, and the launch & subsequent upgrades to higher-priced tiers (family plan and Max).
User growth: Duolingo has sustained impressive growth rates since numbers became publicly available (2019): Monthly active users have grown to 135 million (32% CAGR), and Daily active users have grown to 50 million (48% CAGR).
Paid subscriber penetration: Over time, the company has successfully converted an increasing share of MAUs into paying subscribers, growing from 3% in 2019 to just over 9% today. Of the paying customers; 62% subscribe to Super Duolingo, 29% to the Family plan, and 9% to Duolingo Max. The vast majority (75-90%) of paying subscribers are on the annual plan, which has 2X the lifetime value of monthly subscribers.
ARPU (of paying subscribers) has been relatively steady over time; price increases in penetrated markets (US and Western Europe) and the pull towards higher priced tiers have been offset by consumers shifting from monthly to annual plans and the rollout of lower regional pricing.
Financial Profile
As a result of these factors, Revenue has grown even faster than user growth (56% Cagr). At the same time, gross margins have been stable, just north of 70%. The primary COGS is the take rate of the two app stores: Apple and Alphabet.
The company is oriented around the product: of the 850 employees, 70% are in Engineering, product management, and design roles.
As the business has grown, it has achieved operating leverage through prudent expense and headcount growth: S&M has declined from 21% to 12% of revenue, while G&A has shrunk from 23% to 17.5% of revenue. R&D has followed a similar trend (44% to 30%), but to a lesser extent, as they have continued to invest in product development. As a result, operating margins have swung from -20% to +13%.
The combination of strong revenue growth and operating leverage has led to strong improvements in profitability: pre-tax income has grown to $182 million, Free cash flow is even higher ($370m) as a result of their negative working capital, and ROIC has followed a similar growth trajectory, reaching 52% in 2025. Finally, the company has a strong balance sheet, with $1b in cash and no debt.
Despite 10X revenue growth, a dominant market position, and now meaningful profitability, shares are 19% below their IPO price. Some of this is likely attributable to a reset of an inflated IPO, but the primary reason is multiple contraction on concerns that AI will disrupt the company, more on this later.
In this deep dive, I will set out to answer the 3 key questions I believe are necessary to truly understand the business:
Why has the business been so successful up to this point?
What are the odds this success continues?
Is it available at an attractive price?
1. What has made Duolingo so successful?
Whichever way you look at it: MAUs, DAUs, engagement, revenue, profits, or ROIC, Duolingo is the clear industry leader. Throughout its history, several key factors have enabled this success.
First out of the gate & Tailwinds
In 2011, the app market was growing rapidly but still small compared to web traffic (95%). Rosetta Stone, with its immersion software, was the industry leader while simultaneously transitioning from CD desktop software to a web platform.
At the same time, established language companies were slow to react, falling prey to the innovator’s dilemma; they continued to devote most of their resources to the web product, even as it became clear that mobile was the future. When they did launch mobile apps, they were poorly designed (web-based), slow, and usually required you to be an existing customer of high-priced services, which were at odds with the low prices ($0.99) available on the app store, creating barriers to adoption.
While Duolingo initially launched its product as web-only, the founders quickly realised where traffic was heading and shifted resources immediately. Not only did they pivot quickly, but they also drew design inspiration from mobile-first apps such as Instagram and Path, creating a mobile-first UI. By the time of release, Duolingo was not the first language app, but it was the first good one, and it was free!
Today, 95% of traffic is mobile, and just 5% is web, just as the founders predicted. By offering a best-in-class free product, Duolingo changed the narrative from a nice-to-have supplement to the primary way to learn.
“For the fist time in history, the technology necessary to enable high-quality education is in the hands of billions of people, in the form of a smartphone.”
Over time, the increasing adoption of online language services has expanded the market and taken share away from offline sources, providing a significant tailwind for companies such as Duolingo. By the time Duolingo published its S-1 in 2021, the online segment accounted for $12 billion of the $61 billion language market (20%). Since then, the trend has continued. As of 2026, GMI estimates the language market at $101 billion, with the online segment accounting for approximately 46%.
Gamification approach
During initial development, the two founders came to the realisation that anything can be learnt through a textbook, but few people learn this way because they don’t want to read. Before launching, Luis and Severin tried and failed to learn the courses that the other had created. They figured, if they couldn’t stay engaged in something that they had a vested interest in, why would anyone else? Retention and motivation are the biggest problems for the education industry, and for a subscription service, the key to success is minimising churn (= bad business).
Duolingo is a product-driven company. It is this focus on developing the most engaging user experience above all else that has been the key contributor to their success over time.
While Competition focused on effective teaching (no matter how dry the content), Duolingo made you have fun. They gained 90% of online MAUs, not because they were the best at teaching; they made it enjoyable enough that people came back time after time, forming habits. This counterpositioning is crucial because, unlike school, where students have to sit in class whether they like it or not, the smallest thing can cause a user to close the app or never return. By focusing on engagement first, users learn less on any given day but stick around and learn more over the long term. Duolingo solved better than anyone else the key pain point of self-directed learning: retention.
“To learn a language, you need to practice, and you need to do it consistently over a long period of time. For a user to choose to rehearse sentences in basic Italian rather than scroll through their social media feeds, it needs to be fun and feel rewarding.” (Quatr)
To do this, Duolingo was a pioneer in applying game mechanics to areas outside of the gaming industry. Instead of 30-minute lessons, they are 2-3 minutes long, making learning possible anytime, anywhere. Lessons mixed translation, listening, and speaking into gamified levels, creating a sense of progression.
Over time, the company has constantly iterated and innovated, both big and small, designing a cartoonified UI akin to a game, while adding new features such as the streak, characters, social features (leaderboards, kudos, etc.), and quirky notifications, all of which play on human psychology to deepen engagement and retention. These product enhancements have compounded over time, resulting in a vastly more engaging experience today.
When Duolingo launched, its retention rate was 13% (13% of users came back the next day). The remaining 87% may still return at a later date, but they are not sufficiently engaged to learn. By 2017, this figure had grown to 50% and is around 80% today, more akin to social media retention rates than the 20% industry average.
Today, over 10 million users have streaks longer than a year! As a result of this growing engagement, DAU-MAU, a barometer of user engagement, has grown significantly over time. In turn, this has led to an increasing number of free users upgrading to paid tiers.
Freemium business model
From launch, Duolingo’s product was free, with venture capital funding salaries for almost 5 years, significantly disrupting high-priced incumbents. This lack of monetisation led to constraints: they couldn’t spend on marketing; they had no money; all they could do was build a better, more engaging product. This product-led focus still exists today.
“I am a huge believer that the reason that we have been able to grow so much is because we didn’t monetise early on.” - Luis Von Ahn
Even when investors pushed to monetise the app in 2016, it took 6 months to convince employees that making money was not evil or contrary to their mission to achieve free education for all. In 2017, they launched a freemium model that they still operate today.
The freemium business model is designed for scale: roughly 90% of users still learn for free. By creating a best-in-class product and completely removing the barriers that had previously prevented people (especially beginners) from learning, Duolingo has been able to attract significantly more users than alternative business models. In the US and other English-speaking countries, 80% of new Duolingo users were not learning a language prior to using the app, significantly expanding the market.
As previously mentioned, Language learners have some of the highest churn rates of any industry; roughly 60% of industry-wide users completely abandon their respective apps within the first week. This is much less for Duolingo because the quality of the free product is so good. This means competitors are on a never-ending treadmill to acquire users, typically through performance marketing, resulting in high sales and marketing costs. Peers often spend 35-50% of revenue on sales and marketing, only to see users churn shortly after. As a result of this vicious cycle, companies aggressively monetise existing users (through paywalls) just to break even, capping their potential size.
As Duolingo’s free product has continued to improve with new features and refinements, user numbers, engagement, and retention have all grown rapidly. By 2021, Duolingo had been downloaded 500 million times; the vast majority of which came through word-of-mouth marketing, which is not only the most effective form of marketing (you are far more likely to try something recommended by friends/family) but also comes at no cost! freeing up 35-50% of the revenue competitors spend on acquiring users to further improve the app.
This user-scale and organic-growth engine spreads expenses over a much larger base and generates 10-20X more data than competing apps. As a result, Duolingo has best-in-class economics, providing the data and resources to reinvest more back into the product, widening the advantage of its free offering.
Unhinged marketing
Building on the strength of its word-of-mouth marketing, the company has built a brand synonymous with language learning through a combination of social media virality and efforts to become the standard for language scores. Today, Duolingo is searched 19 times more than “learn Spanish” on Google.
Brand Building: As the company gained momentum, the founders sought a mascot; they chose an owl (named Duo), a symbol of knowledge in many countries. They made the owl green as an inside joke because Severin hated the colour. As Duolingo began experimenting with push notifications (as engagement efforts), the owl took on a now-famous passive-aggressive persona, prompting users to complete their lessons. This relentless and unhinged quest sparked a wave of memes, which made Duo a cultural icon.
Instead of trying to control the narrative, as most corporations would in the same situation, Duolingo leaned into the fun; its social media channels began posting unhinged content aligned with current social media trends. Today, Duolingo is one of the largest brands on TikTok with over 17 million followers! Even more impressively, this growth was entirely organic; they did not spend a single dollar to get there! In 2025, their videos consistently ranked as the week's most viral!
The marketing team has become excellent at inserting the Duolingo brand into cultural moments to stay top of mind. A few examples include:
Their 5-second 2024 Super Bowl ad, which cost $700,000 (one-tenth the cost of a traditional ad), generated 60 million social media impressions.
Their Dead Duo stunt in 2025 drove 1.7 billion organic impressions, while costing virtually nothing.
Additionally, the Duolingo brand (in some form) has appeared in the Barbie movie, on Taylor Swift’s Eras Tour, and featured at Coachella, all of which came to Duolingo.
As a result, ~15% of all new Duolingo users join as a result of Owl-related media. This low-cost virality has likely generated hundreds of millions of dollars in advertising and brand value for the company at little to no cost, resulting in the lowest customer acquisition costs in the industry!
Becoming the industry standard: The next source of organic brand building is to become the global standard for proficiency. Luis Von Ahn wants to make the Duolingo score the measure of language capability. The first step was the Duolingo Test, which launched in 2014 as a cheaper, faster, and more convenient way to assess English proficiency (vs TOEFL and other options). Today, the DET is accepted by over 6,000 education institutions and 99 of the top 100 US universities. They have also been working on acceptance by the UK, Canadian, and Australian Governments for Visa purposes, and have just rolled out integration of the broader Duolingo score into LinkedIn. While the DET accounts for only 6% of revenue, the potential brand and reputation impacts, if it achieves broad adoption, could be profound.
Management & Culture
Management and the culture that has been created are the final contributors to the company’s success. Duolingo did not become the market leader by accident; it is the result of a culture designed to avoid many of the pitfalls that organisations fall into: large, slow-moving organisations with layers of bureaucracy and a lack of accountability. Instead, Duolingo’s mission-driven culture significantly increases the odds of success.
The Long View
To me, the defining characteristic of the culture is its ability to “take the long view”, which is made possible because the company remains founder-led. Luis Von Ahn (CEO) and Severin Hacker (CTO) are committed to making Duolingo their life’s work, protected by a dual-class share structure that insulates them from short-term pressure or hostile takeovers.
“I plan to dedicate my life to building a future in which through technology, every person on this planet has access to the best quality education.” - Luis Von Ahn
Over his career, Luis Von Ahn has repeatedly prioritised the mission over personal wealth; he turned down big tech offers during his time as a professor, today, they purposely leave money on the table by refusing to increase ad loads, and have repeatedly bet on unproven technologies that have the long-term potential to improve current solutions (e.g., Text to speech). The company is unwilling to compromise on achieving its goal of making the best education universally available to meet short-term goals
These actions, along with their continued ownership stakes (~8.5% each) and personal mission, mean management is truly aligned beyond just a paycheck.
Talent density
Over time, Luis has adapted his management style, from micromanaging 50 employees to delegating and trusting as the organisation has grown towards 1,000 employees. However, the one thing that hasn’t changed is an extreme focus on talent.
To this day, Luis and Severin still approve every new hire, a maniacal effort to retain a culture of A-players. Steve Jobs famously said, “A small team of A-players can run rings around a large group of B and C-players.” The company has a saying, “better a hole than an a**hole.”
As a result of its strict hiring discipline and mission-driven culture, the company has a 90% employee retention rate, significantly higher than that of other tech firms. The original group of 30-50 early employees, for the most part, remains at the company, helping maintain the culture.
A culture of testing
Duolingo is built around shipping A/B tests as fast as possible. The green machine is what Duolingo calls this constant iterative process: doubling down on A/B tests that work while moving on from those that don’t. These bets are small and asymmetric. Running tests eliminates the risk of judgment when guessing whether something will succeed; the data tells you.
Over time, the company has run 16,000+ A/B tests; 50% have worked out, demonstrating that it is striking the right balance of ambition. As the number of users increases, the rate at which Duolingo can run these tests accelerates (more users and more data), creating two mutually reinforcing flywheels (learning and investment): more users lead to more A/B tests, which lead to a better, more engaging product, which attracts more users (and paid subscribers), leading to more A/B tests, and so on…This accelerating velocity runs counter to most companies, which slow down as they scale.
As the company scales further, it becomes increasingly difficult to compete with, which has allowed Duolingo to extend its lead.
On the other hand, A/B testing can be dangerous if you don’t prioritise the long-term user experience: the app would end up with heavy ad loads, spammy emails & notifications, which would satisfy the A/B test but are clearly not viable for long-term success. To mitigate this, Duolingo will simply not run certain A/B tests, such as those that increase ad loads. They do this to avoid the temptation to know how much additional revenue they could generate at the expense of the user experience.
Luis von Ahn believes that A/B testing accounts for half of the company's value.
Industry Landscape
Duolingo competes in a highly fragmented industry comprising online and app-based language-learning platforms, as well as offline language-learning options.
The core battleground for Duolingo is amongst competing apps; Duolingo’s 10-K notes that “the online language learning industry is highly competitive, with low switching costs and a consistent stream of new products and entrants.” Over their history, there have been hundreds of new entrants, in over 50 languages, trying to take a piece of the pie.
Despite this, management claims that Duolingo accounts for 85% of all online DAUs and 90% of all MAUs learning online; these figures have been fairly static/increasing in recent years.
The language learning business, on the whole, is not a good business. Duolingo only received VC money because they thought Luis Von Ahn could pivot and sell a third business to Google. Competitors have been unable to disrupt the market or generate satisfactory returns due to the factors already discussed:
Weak retention.
High customer acquisition costs to replenish the leaking bucket, which diverts capital away from product improvements.
Inferior business model choices (paywalls, etc.) due to the need to aggressively monetise the few engaged users
Lack of brand awareness
All of these factors create friction that hinders scaling. While the barriers to entry are low, the barriers to success are high.
Duolingo broke this trend, garnering enough scale to generate attractive economics and ROIC. As a result, many new entrants try to imitate Duolingo; however, these strategies have not worked out so far, likely because Duolingo already exists and running such a business is very difficult. As a result, the greatest threat to Duolingo is not from existing companies gaining market share, who, for the most part, are just trying to keep their heads above water, but from new entrants, technological disruption, and platform shifts.
Zooming out to the broader industry, Duolingo is competing for a share of the 1.5-2 billion people learning a language, putting the company in competition with night classes, private tutors, books, audiobooks, etc. Approximately 7-9% of all language learners are on Duolingo. The future growth opportunity lies in capturing a larger share of the total market. Modor Intelligence expects the online language market to continue to grow by 12-16% per year, providing a tailwind for the company.
2. Odds of continued success
At the beginning of 2025, Duolingo appeared unstoppable, shares peaked at over $520. Fast forward a year, and shares have tumbled 82%, trading around $100. The narrative has shifted dramatically, driven by a series of events.
Leadership
The first controversy occurred in late 2023/early 2024, when Duolingo announced it had laid off 10% of its contract workforce (translators and content creators) because AI could now perform similar roles. This grew into a broader controversy when a leaked memo stated that the company would be AI-first: it would gradually stop using contractors for work that AI could handle, and new hires would only be approved if the work could not be automated. As a result of user backlash (#DeleteDuolingo trended on social media), Luis von Ahn backtracked in May 2025, attempting to distance the company from the AI-first narrative.
Beyond the controversies, there have been two important executive departures: CFO Matt Skaruppa and Global Social Media Manager Zaria Parvez, who was responsible for Duolingo’s unhinged social media presence.
Slowing growth
As a business scales, it is only natural for growth to slow; however, Duolingo is currently facing a slowdown that likely goes beyond the effects of scale.
In Q3 2025, Luis Von Ahn announced a subtle yet significant shift in company strategy. For years, the company has balanced monetisation, user growth, and teaching efficacy; new features targeting one pillar could not be at the detriment of the other two. The new focus shifts the company away from monetisation and towards user growth and teaching better. Management framed this pivot to position the company to capture the huge opportunity that AI presents; Duolingo aims to teach as effectively as a 1-on-1 tutor, expanding its TAM from the beginner/intermediate portion to the entire market (billions of learners).
While AI offers massive potential for the company, and the decision to capitalize on the opportunity is likely correct, the timing is suspicious: bookings growth is slowing, MAUs are flat, and the threat of new competition is looming.
A more likely explanation is overmonetisation. Since going public, Duolingo has leaned heavily into monetisation, rolling out regional pricing, adding a premium tier, raising prices in the most penetrated areas, and increasing the friction of the free tier to convert users into paying subscribers. Additionally, long-time users have been reporting a decline in lesson quality; the novel human-generated stories have been replaced by repetitive AI-generated scripts. The combination of these two factors has led the word-of-mouth growth engine to slow dramatically.
Compounding the problem are changes to TikTok's algorithm, which have made it much more difficult to consistently generate viral content. In 2025, Duolingo had the most viral video of the week on multiple occasions; this virality attracted roughly 20% of new users. Its contribution is likely significantly lower today.
In practice, user growth and learning outcomes will be prioritised in A/B tests, friction will be removed from the free tier (foregoing approximately $50m in bookings), previously premium exclusive features (such as Video call with Lily) will likely be rolled out across all subscription tiers, refocusing personnel on top-of-funnel iniatives, and expanding the marketing engine beyond TikTok. Although these actions will be a headwind to near-term revenue growth, they are necessary. These actions will take time to be felt, but if successful, the user experience will improve, in turn re-accelerating the word-of-mouth growth engine and boosting user growth. Duolingo aims to hit 100m DAUs by 2028, double its current figure.
In the near term, Duolingo expects revenue to slow to +15-18% in 2026, while margins contract slightly due to sustained investments.
The AI threat
As large language models continue to improve rapidly, Investors have begun to ask questions such as: why would anyone spend months or years learning a language when real-time translation is possible? And secondly, why would anyone use Duolingo when ChatGPT is available?
AI is going to change the way people learn - that is likely a fair statement. This creates risks and opportunities that Duolingo must successfully navigate. Currently, there is a significant disconnect between the market’s perception (an existential threat) and management’s view (an opportunity). There are two potential risks to Duolingo: reducing/eliminating the need to learn a language (TAM reduction), and the entrance of new competitors.
Need to learn a language (TAM threat), Google Translate has been around for over a decade, and in most languages it provides almost perfect translation, despite this, Duolingo has grown multiple fold, the problem has been the friction to integrate this technology into everyday situations. New devices such as smart glasses (e.g., Meta Ray-Ban) and wearable technology (like the highly anticipated OpenAI/Jony Ive wearable device) will likely reduce this friction. In the last couple of weeks,s T-Mobile has also announced the availability of live AI translation on phone calls across its network, providing yet another example of AI reducing barriers. As new technologies reduce the friction to incorporate translation, why would anyone spend years learning a language?
My best guess is that to a small proportion of users, real-time translation is enough to discourage them from learning a language. Additionally, certain jobs, such as translators, may be disrupted, leading to fewer job opportunities (reduced incentive to learn). However, to the majority of the 1.5-2 billion language learners, real-time translation misses the point of why they learn a language: to connect with one another. Translation solves an immediate problem; learning a language is a longer-term solution. On the other hand, disruption of any industry begins at the fringe before working its way in, a classic innovator’s dilemma.
Although the technology will continue to improve, conversational translation lacks the natural connection of speech, especially in 1-on-1 settings, and it does not handle overlapping speech or background noise well. Recall watching the news or a Netflix show like Money Heist and how frustrating it is to hear a dubbed-over translation. This is the experience of real-time translation; granted, it is better than using Google Translate, but it is unlikely to ever replicate true understanding.
Looking at the downside, even if end demand is affected, Duolingo has at most a 7-9% market share and is expanding beyond languages, leaving room to grow. On the other hand, Conversational practice is one of the most effective ways to learn a language; these new technologies could inspire a whole new cohort to learn a language, actually expanding the market.
There are plausible arguments both for and against. The simple answer is I do not know whether the total market will grow or contract, but I can conclude with a high degree of confidence that the market will not disappear.
Competition: The second risk is the emergence of new forms of competition. AI is enabling a slew of new competition, from LLMs to new AI-powered apps. While there is currently no indication of any product gaining material market share, the total number of new apps launched on the app store has risen by 60%! Some of which will be competing against Duolingo.
Duolingo did not get caught off guard by AI. Luis Von Ahn was a computer science professor, and Duolingo has been using AI since 2013 (Birdbrain, etc.). They partnered with OpenAI in 2021, prior to the launch of ChatGPT, and were subsequently a launch partner. They were one of the first companies to launch a consumer-facing AI product (Duolingo Max). More recently, there have been two ways Duolingo has been incorporating AI:
Course creation has become increasingly automated to the point where AI now does most of it. This has been cheaper and faster, allowing Duolingo to gather data more quickly while allowing employees to focus on more creative tasks.
New features, most notably conversational practice (Video call with Lily and speaking adventures), data shows that people do not want to practice with another human, but it is one of the most effective learning tools.
There are several factors that make me skeptical that new entrants will outcompete Duolingo.
“Looking at Duolingo can be misleading because the app looks so cute and gamified that many people don’t realise the amount of sophistication that there is in the background and how much personalisation there is in the learning experience.” - Luis Von Ahn.
New entrants have access to the same LLM’s as Duolingo, yet they are lacking the user data, distribution (brand and user base), word of mouth growth engine, and scale that Duolingo already has. Granted, a user can go to any LLM or app to learn a language; however, it takes significant time and practice to make progress. Recall that the biggest challenge for any language-learning app is retention. Duolingo’s key advantage is keeping people engaged through its gamification features, which have been optimised over a decade through 16,000 A/B tests. While competitors may be able to attract users through performance marketing and promises of better results, the real test will be how many will have stuck after a year.
Duolingo’s bulletproof balance sheet insulates the business from the worst-case scenarios. However, on the downside, if competitors gain meaningful scale, Duolingo would likely have to increase investment and marketing spend to compete and attract users, while facing declining retention. As a result, Duolingo’s economics would likely revert to the mean: Language-learning apps do not have great economics overall.
Both threats are real, and mark a point in time similar to the one that allowed Duolingo to disrupt the industry in the first place: while the company possesses strong advantages, it must evolve successfully to achieve similar dominance and economics going forward.
Opportunities
Looking ahead, if Duolingo can navigate the risks above, it has a potentially long runway. In the near to medium term, the majority of growth is likely to come from the core language app; improvements in teaching and engagement could lead to sustained user growth and eventual monetisation. Additionally, the company has the opportunity to monetise free users more effectively through investments in ad tech.
In the long term, expansion beyond languages could drive further growth.
Beyond the top line, Earnings would likely grow faster than revenue due to the operating leverage already discussed in the business.
Teaching Better: Expanding beyond the beginner.
Since its founding, the company has been working to teach and provide private tutoring.
Currently, the app is very effective at getting users to CEFR A2 proficiency (advanced beginner). Independent studies have shown that users who complete 5 units on Duolingo achieve a level equivalent to 4 semesters of university, but can do so in approximately half the time (120 vs 240 hours), however, it still falls short of a 1-on-1 tutor.
Beyond A2, the progression appears to plateau, which, until recently, was all that most courses were designed to achieve, explaining user concentration towards beginners. In the past, the company has had many greenfield opportunities to attract beginner-level users, so this has not mattered; however, as the user base expands, the company must be able to deliver a higher level of learning or risk reaching saturation.
Management’s medium-term goal is to get users to a Duolingo score of 129, equivalent to CEFR level B2, where users can get a job in that language, by harnessing technology to teach as effectively as a private tutor at a fraction of the cost. Recently, the company has placed greater emphasis on achieving better learning outcomes. This focus suggests they are nearing saturation among beginners. It has also been a key barrier in attracting English learners, hence the underpenetration.
In response, Duolingo has released new content in all languages, allowing users to now achieve a Duolingo score of 130 (CEFR B2). They have also launched new lesson types, such as: Adventures, Duo Radio, Stories, Video Call with Lily, and Flashcards. Over the company history, Duolingo, every year, has taught better than the year prior, this trend is likely to accelerate as the company places a greater emphasis on it.
Teaching better ultimately translates into user growth (with a lag) and, subsequently, greater monetisation (a much larger lag). From the user’s perspective, a course that teaches better will take time to be felt. Only then, once the user has realised they have made progress, will they begin spreading the word.
Importantly, Duolingo must tread a fine line: they cannot improve learning so much that they impact engagement, which may anchor them and hold them back from attracting the most advanced learners over time.
Engagement: Engaging maintenance users.
The company does not disclose average time spent, duration, etc.; however, Screenshots on Reddit forums from end-of-year statistics suggest the average time spent is in the low single-digit minutes-per-day range, with the 0.1% most engaged users typically spending up to 45 minutes on the app. The reality is that while many users are classified as DAUs, they are not truly engaged; instead, most complete only a single lesson per day to keep their streak going.
While some may consider this a risk, management acknowledges that a significant chunk of users return only to keep the streak going (maintenance users), which ties back to the key issue of self-learning: motivation. Additionally, these maintenance users are unlikely to be paying users, meaning they contribute very little to the financial picture.
I suspect that the low time spent on Duolingo is a feature of the industry rather than a bug in Duolingo. These users would simply drop off competing apps, never showing up in the numbers, but Duolingo keeps them coming back, even if it is for one lesson, elongating the engagement profile, but providing useful data and multiple opportunities to engage these users further.
To me, the crucial factor is not the absolute time spent, but the trend: is it increasing or decreasing? Several factors give me confidence that engagement is heading in the right direction, albeit off a low base:
The number of lessons completed per day has grown from 500m (2021) to 900m (2022) to over 1B (reported in 2023 and 2024) to nearly 2B (2025), likely crossing 2B imminently.
The DAU-MAU ratio has increased from 20% a few years ago to 37% today, indicating that users, on the whole, are coming back more frequently.
Management has consistently reported an increase in the median time spent after product feature enhancements.
If the company can continue to teach better and improve engagement, retention will increase, and user growth will accelerate (via word of mouth and attracting more advanced users), benefiting both ad and subscription revenue.
Combined, better learning outcomes and improved engagement should allow the business to continue attracting more users. Duolingo currently has 7-9% of language learners (at most), indicating plenty of opportunities ahead.
Monetisation: Optimising for the long term.
Although revenue growth is likely to be slower over the next couple of years due to shifting priorities, the company has several opportunities over the longer term to increase overall monetisation beyond the benefits of user growth mentioned above.
Paid Penetration: The number of paid users has grown rapidly since going public (54% Cagr). Despite this, the app remains meaningfully underpenetrated (9%) compared with other freemium mobile apps: Match Group (18%) and Spotify (38%). Over time, I don’t think it’s unreasonable to believe that Duolingo could match or surpass the penetration of dating apps, if it can attract intermediate/advanced users who are more willing to pay. However, given the immediate change of focus, this is likely to be a longer-term goal.
On the other hand, if they are unable to attract more serious learners, Duolingo’s penetration within the casual learners segment, who view it as low-stakes entertainment rather than essential education, could create a ceiling to monetisation.
Ad tech: Another undermonetised area is the advertising business. Duolingo has historically shown a single programmatic ad from Meta and Google’s ad platforms after every completed lesson, typically featuring low-quality ads for other mobile games. While the ad load is unlikely to increase, taking control of the ad platform could result in significantly higher revenue per ad. The company has begun taking steps towards this goal, launching its own ad platform with direct deals for major brands like Disney. Longer-term, Von Ahn has described his dream of showing ads partially in the language being learned, the idea here is that using it as a learning experience, users would actually be more engaged with the ad. It is unclear whether this will work out; however, at only 8% of total revenue, the company has a significant amount of low-hanging fruit that should allow the company to better monetise the 120m+ free users on the platform, which is crucial for countries such as India that have a lower propensity for subscriptions.
Beyond Language
Duolingo’s mission statement does not mention languages.
What was solely a language app, the company has gradually expanded into additional subjects. The Maths course came first, initially launched as a standalone app, before being rolled into the flagship app with the launch of Music. These subjects were chosen first because billions of people want/need to learn these subjects. In Q4 2024, management reported that the two courses had a combined 3 million DAUs.
Chess was the third subject launched in mid-2025, taking 2 employees (with no coding or chess experience), 9 months from idea to launch! It has since become the fastest-growing subject in company history, and the second-largest chess platform in the world, reaching 7 million DAUs within a year. Although other platforms have shown that multiple subjects are viable, chess's success further confirms that Duolingo can expand beyond languages.
At the moment, these subjects do not have a paid tier, but do show ads. Similar to Meta (but on a smaller scale), the company will let a new product scale organically before even considering monetisation. Management is taking a disciplined approach to expansion, making sure they nail the current syllabus before expanding. Currently, no new subjects are being developed, but in the longer term, it is highly likely that more subjects will follow, such as Science and other strategy games that follow in chess's footsteps. Ultimately, new subjects must have a huge demand (hundreds of millions of people), be good for the world, take a long time to learn, and be teachable through a mobile app.
Subject expansion, if successful, could extend the TAM from the online language industry to the $300 billion online education industry, and, as with languages, Duolingo could expand into these individual markets; however, meaningful benefit is likely still years away. Framing it another way, Khan Academy has over 190 million registered users on its platform; there is no reason why Duolingo could not achieve a similar scale outside of languages.
Other risks
Beyond the main risks mentioned earlier (disruption), there are several other risks, such as app store reliance, the impact of any macro downturns on users’ willingness to pay, and algorithm changes (social media platforms) impacting the viral marketing engine. Although these are all real risks, they are not isolated to Duolingo; every company must navigate them.
Stock-based compensation, while not a risk to the business's success, is a significant risk to shareholder returns. Since going public, revenue has compounded at 56%; however, revenue per share has compounded at less than half that rate (24.6% CAGR) due to significant dilution. This is due to shares outstanding increasing from 12 to 48 million since 2019 (25% Cagr). Although dilution has been improving (13% SBC-to-revenue), it remains meaningful and is expected to increase in 2026 as the company invests in growth. As a result, the company must continue to deliver outsized growth to achieve satisfactory incremental earnings per share (re-acceleration); otherwise, per-share value will remain stagnant.
Capital allocation is also a question mark; throughout the company’s history, they have devoted all available resources to reinvesting in the business, which has been and for the moment will continue to be the right thing to do. However, as the company begins to spit out cash, capital allocation is becoming an increasingly important consideration. Over the last couple of years, the company has acquired a few small animation studios and has recently announced a $400m share buyback program; however, most of this will merely offset dilution. While I do not know how they will allocate capital going forward, the company’s intense focus on ROI throughout, frugality with operating expenses, and alignment with shareholders give me comfort that they will not waste cash.
3. Valuation
Following an 80% drawdown, Duolingo is currently trading at its lowest ever valuation: $3.7B EV, 4.6X sales, or 12.8X FCF. The market, as a result of disruption fears and shifting priorities, is likely pricing in approximately 7% FCF/ share growth going forward.
If investors believe the company can deliver outcomes superior to market expectations by successfully navigating risks and executing on its growth drivers, then the currently low expectations may present an attractive opportunity.
In my view, the company represents an asymmetric opportunity:
Looking at the downside, the business has over $1B of cash and no debt on the balance sheet, a highly scalable and attractive business model, and multiple advantages (scale: users + data, brand, and founder-led culture) in an industry where scale is the only way to thrive (low barriers to entry but high barriers to success). These factors reduce the downside possibility. However, the looming threat of disruption and competition cannot be discarded.
Additionally, industry tailwinds (from offline to online learning), improving teaching outcomes and engagement, and expansion beyond languages provide a long runway for user growth. This user growth, coupled with increased monetisation, provides the possibility for significant growth.
Conclusion
Duolingo is a fascinating company; it has achieved 80% market share in the online language-learning industry as a result of its best-in-class gamified product, viral marketing engine, and long-term data-driven culture.
Despite significant upside potential, Shifts in priority and the threat of disruption have worried the market, potentially creating an attractive entry point or an existential threat. Before investing, investors must be comfortable trusting Luis Von Ahn to navigate AI risks and a near-term slowdown well enough to more than offset the material stock-based compensation.
Sources: Already linked content above, Company material, Quatr, Duolingo Handbook, Decoder with Nilay Patel, ACQ2, The Eric Ries Show, View From the Top, The Time Ferris Show.
Disclaimer: Disclosure: I/we may or may not have a beneficial long position in any of the securities discussed in this post, either through stock ownership, options, or other derivatives. This article expresses our own opinions. No recommendation or advice is being given as to whether any investment is suitable for a particular investor. The information provided in this analysis is for educational and informational purposes only. It is not intended as and should not be considered investment advice or a recommendation to buy or sell any security.
Investing in stocks and securities involves risks, and past performance is not indicative of future results. Readers are advised to conduct their own research before making any investment
Hey here. Great content. I will read all your deep dives. I have subbed. Let’s grow together and support each other. Check my deep dive on CRM here https://substack.com/@valueinvestorfromitaly/note/p-190818961?utm_source=notes-share-action&r=3qdo3i
DUOL could generate significantly more revenue from blended-in ads across all tiers. These kinds of ads wouldn’t hurt the Duolingo experience, especially if they partner with well-known brands and integrate them naturally into lessons. Hopefully, they implement this soon, as it could drive Duolingo’s growth quite easily without requiring major improvements to the core business