Keeping the Streak Alive: The Story of Duolingo

Even if you've never used Duolingo, you've more than certainly heard of it. Maybe it's a friend bragging about their streak, or the unmistakable Duo owl in your social media feed. But behind the neon-green app interface sits a data-driven education company, where every design choice, feature, and push notification is carefully engineered from the experience of hundreds of billions of lessons. This is the story of how Duolingo turned language learning into a game – and a business.

A life shaped by learning

Our story begins far away from Silicon Valley. Instead, we're heading to our protagonist, Luis von Ahn's hometown of Guatemala City. Born at the end of the 1970s, he was raised by his mother and grandmother during a period fraught with economic hardships while a civil war between the government and various rebel groups raged.

Despite the family's limited means, Luis von Ahn's mother ensured that he received a proper education. In a 2022 interview with ABC, he described how his mother did everything she could to secure him the best possible education: “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.”

While education would be one of the pillars shaping von Ahn's path through life, a Commodore 64 would be the second. His mother had chosen not to get him a Nintendo, instead opting to get him a computer, much to the dismay of young Luis. But it would turn out that his mother had made a great decision when choosing a gift. Quickly brushing off the disappointment, von Ahn got to learn the Commodore 64 inside and out, and soon, he was coding. That choice of gift set in motion a journey he's still on today.

Before doctoral degrees and Duolingo, there was a Commodore 64.

Two exits to Google

After completing high school in Guatemala, von Ahn applied for university studies in the U.S., and thanks to his educational background, he earned a scholarship at Duke University. Following undergraduate studies, he continued in the path of academia, moving on to pursue a Ph.D. in Computer Science at Carnegie Mellon University.

In one early project at CMU, he created an online collaborative game for image labeling, which paired random users to assign descriptions to pictures. Aside from being fun for the players, it also generated valuable data for a reverse image search engine. In 2006, just a year after von Ahn had earned his Ph.D., Google acquired the project for around $2 million.

During his years at university, von Ahn also co-developed the concept of CAPTCHAs. While you might not recognize the name, you've almost certainly seen them in action: it's those distorted text images used to tell humans and bots apart online.

However, von Ahn didn't like the fact that hundreds of thousands of hours were being spent each day on typing in CAPTCHA words with no further value. So, he continued to build on his co-innovation, developing a twist on the idea where each prompt wasn't just a security test but also a snippet from a scanned book that needed transcription, and called it reCAPTCHA. In effect, people entering text to prove they were human were simultaneously helping to digitize old books for the Internet Archive. Once again, Google took notice and acquired reCAPTCHA in 2009.

At just 30 years old, von Ahn had two successful exits, and offers from big tech companies were starting to pile up. Regardless, he stayed in Pittsburgh as a professor at Carnegie Mellon, continuing to teach his classes in Computer Science.

Creating Duolingo

In the early 2010s, digital learning was a budding industry, and several platforms for coding and maths were starting to pop up. While there were plenty of early imperfections that would be smoothed out over the years to come, the concept clearly had potential. Teaching a small class of computer science students certainly had its appeal, but digital learning provided an opportunity to reach millions. Luis von Ahn, by now growing a bit restless in academia, felt the pull to do something bigger.

So he started formulating a plan. In search of a partner he began looking among his students for someone young, ambitious, and technically skilled. He found that person in Severin Hacker, a Swiss Computer Science Ph.D. student with a somewhat ironic last name who shared von Ahn's passion for using technology to solve real-world problems. Together, they were going to build an educational tool, that much was certain. The only thing left to figure out was what to teach.

Hacker was initially leaning towards programming or math, subjects they both knew like the back of their hand. But von Ahn steered the conversation into a completely different direction: languages. He was deeply influenced by his own life experience, as learning English had unlocked opportunities for him that otherwise would have been nothing more than a pipe dream. Enabling anyone, anywhere, to learn English for free would be a complete game-changer for billions of people.

Launch and rapid growth

Luis von Ahn and Severin Hacker founded Duolingo in 2011. Initial funding came somewhat unconventionally as von Ahn used part of a grant he'd been given, and also secured a second grant from the National Science Foundation to get started. By late 2011, the project had attracted investors for a Series A round, and the small team, comprised of von Ahn, Hacker, and a handful of others, could start building.

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.

After about a year of beta testing, Duolingo officially launched to the public in June 2012. It arrived first as a website but was quickly released as a mobile app, with English and a handful of other languages available. Just like von Ahn had predicted, the interest was huge.

Following the release, things moved at lightspeed. Usage quickly climbed into the millions, course offerings expanded, Apple named it App of the Year in 2013, and as it launched on Android, it further expanded its reach. By the end of 2013, Duolingo had over 10 million users.

Inside an early Duolingo office, with the team hard at work on the app.

Rethinking monetization

Potentially helping billions to learn a new language was an admirable idea. But goodwill does not pay for salaries and server costs. Here, von Ahn's earlier work inspired a potential solution. Just as reCAPTCHA had repurposed human effort, the duo conceived a model where users learning a language would simultaneously help translate real-world texts. In theory, it could generate revenue by providing translation services, without charging users a single penny.

During these early years, that crowdsourced translation idea was put to the test. Duolingo's business model 1.0, or the “immersion” translation platform, went live alongside the language courses. As learners progressed to intermediate levels, the app would present real news articles or Wikipedia entries for translation practice. Users translated sentences into the language they were learning, and these translations were voted on and refined by the community for accuracy. But as great as it sounded in principle, it brought with it several issues.

As 2013 drew to a close, Duolingo shut down the immersion translation platform. While the idea seemed promising, the data told a different story. The feature added unnecessary complexity and didn't improve the core learning experience. Moreover, relying on translation by learners to fund the whole enterprise felt limiting. They needed users to enjoy the app, not feel like unpaid translators.

Duolingo continued to raise capital in 2012 and 2014, thereby easing short-term monetization pressure and allowing the company to focus on building a long-term product. In fact, it would prove to be the case that Duolingo didn't generate meaningful revenue until 2017. There was no marketing spend at all, and none of the monetization features that define Duolingo today: no in-app purchases, no ads, and no subscription tiers.

In a 2024 interview on Acquired, Luis von Ahn said that: “I am a huge believer that the reason that we have been able to grow so much is because we didn't monetize early on.” That deliberate strategic choice forced a product-first focus:

Of course, there's a degree of survivorship bias here. Doubling down on organic growth and long-term product quality looks obvious in hindsight, but plenty of apps have tried the same playbook and failed. But Duolingo didn't. The choice to optimize for users rather than short-term revenue has turned out to be a very wise decision.

Designing for engagement

Free from the translation experiment and with a solid runway, Duolingo's developers focused on refining both the gamification and the learning experience. In practice, the two are inseparable. 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.

By introducing things like cute animations, encouraging sounds, competitive leaderboards, and the now-famous streak count (which tells you how many days in a row you've practiced), Duolingo ramped up its efforts to hook learners. By making language learning feel more like a game than being back in middle school Spanish class, Duolingo had solved a problem that had plagued digital learning for as long as the medium had been around: retention.

Boosted by these additions, Duolingo's user metrics continued to climb exponentially. By mid-2015, monthly active users (MAU) soared into the tens of millions, while the app reached over 100 million registered users since launch – extremely impressive, given that it was an education app. The secret behind all of this was an obsession with driving engagement and retention through constantly tuning its content and features through A/B testing. Duolingo has, from the start, had an extremely data-driven culture. Every single change – from the color of a button to the phrasing and timing of a notification – was (and still is) tested on small groups to see its impact on retention and learning before being implemented for all users.

The result of this relentless experimentation and optimization kept users coming back, day after day, week after week, and soon enough, year after year. Users wanted to maintain streaks, complete their daily goals, and not let down Duolingo's owl mascot. The green bird, called Duo, was starting to take on a life of its own during this period and has become a massive asset for Duolingo's marketing department in recent years.

Duo was (and still is) used as a clever way to drive engagement. Whether it was sending out a push notification about how sad Duo was that a user hadn't completed their daily Polish lessons, or posting a video about how he'd visit you at night (wielding a baseball bat) if you didn't do your French word matching exercises, it worked. People kept coming back to keep their streaks alive, learning more and more day by day.

Duolingo's push notifications, a key mechanism for bringing learners back to the app.

Turning growth into revenue

Duolingo's user growth and retention numbers continued to be strong by the mid-2010s, but MAU numbers in and of themselves don't keep the lights on. During these years, the company had begun creating minor sources of revenue by enabling some ads on the platform and launching its own English Proficiency Test. But in 2017, it took a large step towards proper monetization by launching a subscription tier: Duolingo Plus. We'll go into all of these additions in detail later.

The important thing to take away from this is the fact that monetization was implemented in a user-friendly way, consistent with Duolingo's mission of making education accessible to everyone. The subscription tier would help keep the app free, while users who were willing to pay a few dollars per month received a somewhat more convenient experience with some extra features.

With revenue on the climb and continued momentum in user metrics, Duolingo secured additional funding in 2017, valuing it at $700 million. Growth and valuation continued to climb from there, and two years later, a funding round led by Google Capital brought the company's valuation to $1.5 billion. Duolingo had officially become a tech unicorn, all while fundamentally remaining an education company with a social mission. But Luis von Ahn had no intention of slowing down.

Going public

By the turn of the decade, Duolingo was, by far, the most popular language learning platform globally, and was being used by both the poorest and the richest people in the world. At the start of 2020, the company reported that it had over 300 million registered users. Active user counts were just as impressive, with MAU and DAU (daily active users) reaching 30 million and 5 million, respectively. Growth got another jolt as the pandemic forced people to stay at home, and millions turned to online learning to pass the time.

After a decade of growth, Duolingo went public in the summer of 2021 at a valuation of $3.7 billion. The IPO was a resounding success, and at the end of the first trading day, Duolingo's market cap had jumped to $5 billion. Luis von Ahn and Severin Hacker (who still remain CEO and CTO after all of these years) had proved that a socially conscious startup driven by something more than just making money could be sustainable, and in due time, profitable.

Inside the Duolingo app

Before we get into Duolingo's business of 2025, we'll give you a short rundown of the in-app experience. Using Duolingo is intentionally simple from the moment you open the app, and not how one would traditionally think of an educational platform.

How Duolingo's in-app learning experience has evolved.

The app is free to download, free to start, and free to use indefinitely in any country in the world. You can learn a language from start to finish without paying a cent. Its courses now span around 40 languages, but the reality is that most learners gravitate toward the same top languages, where demand concentrates heavily. The fact that 80% of language learners in the world are learning English gives you a good sense of what demand looks like.

Lessons are short by design, built to fit into small pockets of time. Each one mixes translation, listening, speaking, and multiple-choice tasks, wrapped in a progression system of levels, checkpoints, and guided by its characters. The structure gives learners a sense of constant forward motion, and it is reinforced by the streak mechanic. We'll get a chance to revisit the streak later in the lens of conversion.

Monetization now comes from different streams. There's advertising for free users, and in-app purchases offer one-time perks like streak freezes or time boosts. But the real revenue driver comes from subscriptions: Super Duolingo centers on convenience and a smoother experience, removing ads and unlocking extras, while Duolingo Max takes it further with AI-powered tools such as conversation practice.

As mentioned earlier, the company is relentless in its pushes. These are intentionally designed and timed to keep you engaged. If you've lost your habit and your streak, the clever push notifications might be what “resurrects” you. Von Ahn said on the Acquired episode:

Beyond language learning

For its first decade, Duolingo was dedicated solely to language learning. But during the 2020s, it has broadened. Now it can fairly be described as a learning universe, one that has expanded beyond languages into other skills. In 2022, Duolingo launched a standalone math app, followed by a music course in 2023. Both were integrated into the main Duolingo app later in 2023. Chess followed these expansions in 2025, first announced in April and launched in beta on iOS in mid-May, and has so far been "growing much faster than math and music and faster than the way originally languages grew."

At the Q1 2025 earnings call (sourced through Quartr Pro), von Ahn articulated Duolingo's expansion strategy clearly:

When asked about future subjects at the Q3 2025 earnings call, he was direct: "at the moment, we're not working on any other subjects." The focus is instead on deepening and improving what already exists, expanding math to cover all K-12 common core content, revamping the music course, and refining chess with features like player-versus-player mode. However, while it is now progressing slowly with adding additional subjects, he laid out a longer vision for Duolingo's role in education: "... it's going to be years until we get to a point where we have an app that I think is just the best possible way to learn any major subjects."

Duolingo's English Test

Alongside its learning courses, Duolingo operates another business: certifying English proficiency. Launched in 2016, the Duolingo English Test (DET) is an alternative to established exams like TOEFL and IELTS – tests that international students take to prove their English language skills when applying to universities in English-speaking countries.

The Duolingo version was built to be more accessible and convenient while still meeting the standards universities require for admission tests. The DET is not completed through the app but from a computer and requires no more than a webcam and an internet connection. Anyone can take it from any location for $49, which is a fraction of what traditional proficiency tests charge at physical testing centers, and delivers the results within days.

At the end of 2021, over 3,000 higher education programs accepted the test. By Q3 2025, acceptance had crossed 6,000 institutions, including all Ivy League universities and 99 of the top 100 U.S. universities. While the DET generates a little more than 5% of company revenue, its significance might as well run deeper than that line item. Duolingo is no longer just an annoying owl on your phone, but a recognized part of educational systems around the world.

Building the Duolingo brand

Much of Duolingo's success can be explained through product and engagement mechanics. But a considerable part of its rise in popularity comes from the brand. No other education company has ever built something so culturally loud, self-aware, and instantly recognizable. In the Acquired interview, Luis von Ahn estimated that roughly 15% of new users arrive because of “some sort of owl thing.” That “weird stuff” may sound like a fluke, but it is the output of a marketing team that understands the internet better than most consumer brands.

The company's social media presence has become a case study in how to turn a software product into a meme engine. Duolingo's team embraces absurdity, parody, and manufactured chaos to spark conversation and cultural relevance.

One viral campaign, “Duo Owl fakes his own death,” generated tens of billions of impressions worldwide. In the stunt, Duo appeared lifeless on social media and users were told they could “bring him back” by doing lessons. The whole thing was surreal, deeply on-brand, and instantly recognizable as Duolingo. This tone also shapes how Duolingo communicates inside the product.

The brand gives the company permission to be more direct, more playful, and more aggressive than a traditional education platform ever could be. That includes the lightly guilt-driven nudges and the occasional passive-aggressive reminder. Along with Duo, the company has created several other characters, all contributing to a more engaging learning experience and boosting retention. All of these features tie back to the same idea that learning requires consistency. We'll get back to the mechanics behind habit-building and retention, but the brand is what makes those mechanics feel entertaining rather than clinical.

Perhaps the only app bold enough to turn its mascot’s ‘death’ into a marketing campaign.

Learner ambassadors

While Duo captures attention, its users are the ones who spread the app most effectively. Anyone who knows a devoted Duolingo learner has probably heard about their Norwegian streak (likely too many times). That constant, almost ritualistic engagement creates millions of informal ambassadors who market the product simply by using it.

All of the lessons completed every day on commutes, during lunch breaks, or while winding down at night total over one billion. The commitment to those sessions is practice for the user and promotion for Duolingo. People will let you know that they have to practice, talk about their streaks, share milestones, brag about leaderboards, and compare progress with friends or family. A single learner can pull several others into the ecosystem without Duolingo spending a cent.

As a result, one of Duolingo's most effective acquisition channels is everyday behavior. The more the app becomes part of someone's routine, the more likely they are to talk about it, and the more the cycle repeats.

Education, game, or both?

"It's not a language learning app, it's a game" must be one of the most common critiques leveled at Duolingo. Some skepticism is expected when a company attempts to capitalize on educational tools, particularly one that leans so heavily into gamification.

But gamification, buzzword though it may be, isn't inherently problematic. The question isn't whether Duolingo uses game mechanics but whether those mechanics serve learning or distract from it. Because if gamification is the best means to an end, it can be extraordinarily effective, and the two don't have to be mutually exclusive. Compared to many other educational apps, Duolingo has found a rare balance. Users don't see it as a stiff, formal learning platform that feels like homework. That's the key to getting millions of users to consistently engage without being required to, on their own free time, day after day. Luis von Ahn addressed this directly in the 2024 Acquired interview:

That insight cuts to the core of Duolingo's strategy. The company has effectively turned language learning into productive screen time. Duolingo is a game. But it's a game designed with a purpose: to make learning something people actually want to do. And if users are learning languages while playing, then perhaps the distinction matters less than the outcome.

Duolingo and AI

In April 2025, Luis von Ahn posted an all-hands email on LinkedIn making something official that had been implicit for a while: Duolingo is going to be AI-first. In the communication, he argued that “AI isn't just a productivity boost”, but with it, Duolingo gets closer to its mission of teaching everyone. In the message, he stressed that AI isn't for replacing “Duos.” Instead, the plan is to remove bottlenecks so employees can spend more time on creative work. To reinforce that, he laid out constraints: gradually move contractor work to AI where it makes sense, make AI literacy part of hiring and performance, and only add headcount where work genuinely can't be automated.

However, Duolingo's integration of AI began much earlier. As said by Luis von Ahn at its Q2 2023 earnings call, it had “been using artificial intelligence for years.” In its 2021 annual report, its first as a public company, that push was a feature in its operations:

That 2021 description still captures the core of Duolingo's AI advantage. Since then, the daily volume of exercises has grown from over half a billion to well over a billion, feeding into its AI models and powering decisions across the entire company. Duolingo's work with AI in 2025 consists of multiple layers that span content creation, personalization, and operational efficiency.

Beginning with the learning side, a major part of this is that it helps Duolingo ramp up the scale of its content creation. In its first 12 years as a company, it added roughly 100 language courses. In 2024 alone, the company added 148 new language courses. This applies to both adding new languages and entirely new educational verticals, but also increasing the offer for its current languages.

Additionally, AI powers entirely new product capabilities that were simply impossible before. Video Call with Lily, Duolingo's conversational practice feature within Duolingo Max, perfectly exemplifies this. It allows learners to practice real-time conversation with an AI character that adapts to their level and corrects mistakes instantly, without ever running out of patience.

On the engagement side, with its A/B testing, AI helps process the massive amounts of behavioral data to identify patterns and personalize the learning path for each user. The relationship between data, AI, and product improvement creates a powerful flywheel effect: more users generate more learning data, which trains better AI models, which enable better personalization and new features, which attract more users and increase engagement. In the end, that leads to more subscribers and improved monetization.

As we'll see later, that context is crucial for understanding why, in Q3 2025, Duolingo decided to tilt even more focus toward these areas.

The threat of AI

Ever since ChatGPT launched in late 2022, AI has been a constant discussion topic related to Duolingo. The questions tend to fall into two buckets: will general-purpose AI tools like ChatGPT, with its education-focused GPTs, or Google's Gemini, with experiments such as Little Language Lessons, replace Duolingo as a place to learn? And will ever-better real-time translation make learning a language unnecessary in the first place?

On the first point, Luis von Ahn addressed the threat directly during the Q3 2025 earnings call. Regarding ChatGPT as a competing learning tool, he said:

This touches the core of Duolingo's competitive advantage. Its models are trained and refined on hundreds of billions of lessons completed inside Duolingo's specific curriculum and interaction patterns. In an AI-powered world where access to general-purpose models is increasingly commoditized, ownership of rich, domain-specific datasets becomes the real moat.

A large language model can be incredibly competent at generating examples or explanations in theory, but without years of structured, labeled learning behavior – when people drop off, what they get wrong, which sequences keep them engaged – it lacks the empirical map of how real learners actually progress. Duolingo owns that map, and every additional lesson completed on the platform makes it more detailed, more predictive, and harder for others to copy.

On top of that data advantage, Duolingo is also ahead in turning all of this into sustained motivation. Building a model that explains a grammar rule is comparatively easy; building a system that keeps tens of millions of people coming back to practice, day after day and year after year, is not. Other language-learning apps like Babbel and Rosetta Stone remain part of the competitive landscape, but they operate with far lower scale.

The second fear: that AI will make language learning obsolete, was also addressed by von Ahn in the same call:

The nuance behind this argument is that translation solves a transactional problem, not an aspirational one. People learn languages for joy, culture, or travel, because it expands their options in education or work, or because it helps bridge the gap to someone they want to get closer to. For all of those reasons, relying on a phone as a permanent intermediary is neither practical nor desirable.

The business beneath the app

For 2025 Q3 TTM, Duolingo generated $964M in total revenue. Around 81% of that now comes from subscriptions, with the remaining share from advertising, the Duolingo English Test, and in-app purchases. The subscription mix has steadily increased from the low-70% range in 2020, growing revenue at a 52% CAGR over the period.

Starting in 2024, Duolingo began reporting advertising, the Duolingo English Test, and in-app purchases together under a single “Other” category. The reporting structure reflects where the real focus lies: the subscription business.

Freemium at scale

Duolingo operates on a freemium model that pairs broad free access with premium monetization. Over time, this structure has been refined to maximize both reach and revenue. While 90% of learners use the platform for free, the company has built a sophisticated subscription engine that focuses on converting engaged users into paying customers. The key to the model is the size and growth of the free-user top-of-funnel, where deeper engagement is nurtured over time. Only a small share of these learners eventually convert, but those who do tend to stay for the long term, making each subscriber highly profitable.

After introducing Duolingo Plus in 2017, the company steadily broadened and improved its paid offering. The plan, rebranded as Super Duolingo in 2022, is the entry tier, priced at around $12.99 per month or roughly $5 per month when billed annually. Super remains the backbone of the subscriber base and accounts for more than half of total subscription revenue.

The Super Duolingo Family Plan is an annual, upfront subscription at $119.99 for up to six users. Its popularity has grown consistently since launch, and it now represents 29% of the subscriber total.

Above these sits Duolingo Max, launched in 2023. Max adds AI-powered features such as Explain My Answer and Video Call, which allow for conversational practice with an AI character. It's priced at about $29.99 monthly or $14 per month on an annual plan, with a family option at $240 per year. Since launch, Max has grown to roughly 9% of subscription revenue. While Max still represents a smaller share of total subscribers, its higher price point makes it an increasingly meaningful portion of subscription revenue. And because of its features, it makes it a very promising tier in the years to follow:

With more than 80% of total revenue coming from subscriptions, conversion is central to the business. Duolingo's extensive A/B testing refines its paywall strategy: adjusting ad frequency, timing, placement, and the user-specific thresholds that determine when someone is likely to convert. As we'll discuss later, this exists within a larger balancing act. Push the levers too hard and you can harm engagement. Pull them too lightly and you leave revenue on the table.

Still, the company's strongest conversion tool isn't a paywall prompt or an ad. The design of the app, the habits it builds, and the engagement loops it creates do much of the work rather than an offer screen. That dynamic comes into sharper focus when looking at the company's engagement metrics.

Enduing engagement

Across its plans and tiers, Duolingo has 11.5 million subscribers as of Q3 2025, up 34% year-over-year. The company's two core operating metrics are MAU and DAU, which together capture the improving engagement and stickiness of the app. In Q3 2025, MAU reached 135.3 million, up 20% year-over-year, while DAU reached 50.5 million, up 36%.

The DAU/MAU ratio measures how frequently users return. Its trend over time gives a sense of whether engagement is strengthening or weakening and, by extension, how large the pool of potential subscribers is becoming. Duolingo has steadily improved this ratio, as seen in the image below. In Q3 2025, it reached 37.3%, up from 32.9% a year earlier.

The fact that this has consistently risen while the user base has scaled so quickly only makes the improvement more notable. Each percentage point reflects a reinforcing loop of data-driven decisions sharpening learning and gamification, combined with distinctive marketing that keeps learners coming back.

Duolingo's DAU/MAU trend, showcasing growing user stickiness.

Share

Another key driver of engagement is the streak mechanic. As of December 31, 2024, roughly 10 million users had streaks longer than one year, highlighting just how deeply the app has embedded itself into daily routines. Duolingo notes that many users spend months or even years on the app before subscribing.

The streak system creates a sense that missing a day of Portuguese would waste all the effort already invested. That tension between not wanting to lose progress and wanting a smoother experience on your daily return helps nudge learners toward paying. Von Ahn put it simply on the Q2 2024 earnings call:

Balancing strategy and finding the perfect mix

During Duolingo's first five years, the goal across the organization was straightforward: grow users and improve how well the app teaches. That meant building the most engaging product possible for learners, without yet having to weigh that against monetization or other trade-offs.

Over time, as Duolingo began generating meaningful revenue, went public, and eventually turned profitable, that optimization problem became more complex. Ads, subscriber conversion across multiple tiers, and in-app purchases all became important levers. Each of these could drive revenue, but each also had to be weighed against user engagement, growth, and learning outcomes. What emerged was a constant trade-off and a very delicate balancing act. Luis von Ahn described it on Duolingo's Q4 2022 earnings call:

Ever since it began monetizing the app in 2017, this framework has worked incredibly well. Monetization, engagement, and teaching efficacy have all improved together, and the guardrails around each team's objectives have prevented any one priority from overwhelming the others.

All of this ultimately rolls up into a single target: platform lifetime value (LTV), the total value a user generates over their entire relationship with Duolingo. Platform LTV sits at the core of Duolingo's strategic decision-making. It's a composite of three pillars – engagement, teaching efficacy, and monetization – each operating on a different timeline but with the same objective.

Monetization improvements show up the fastest, immediately lifting bookings through better pricing, conversion, or increasing advertising. Engagement improvements take longer as features that make the app more fun or habit-forming first show up in metrics like DAU, streak length, and session frequency before it translates into more subscribers. Teaching improvements take the longest. Better courses and explanations initially drive learning outcomes, leading to increasing engagement and retention, which in turn support user growth and, only eventually, monetization.

The balancing act is about maximizing the combined long-term value of these forces. Some decisions may slow growth or monetization in the short term, but are designed to increase LTV over a longer horizon. Being founder-led, with management explicitly aligned around that long-term optimization, has likely made it easier for Duolingo to stick to this approach.

Growth and conversion

Everything we've covered so far, the massive free-user base, the habit-building product, and the careful balancing of teaching, engagement, and monetization, shows up most clearly in Duolingo's user growth and conversion.

As long as the base of free learners keeps expanding, Duolingo's pool of potential subscribers grows with it. Over the past few years, that dynamic has been on full display. From Q3 2020 to Q3 2025, MAU increased from 37 million to 135.3 million. Over the same period, DAU rose from 8.4 million to 50.5 million, and paid subscribers climbed from 1.5 million to 11.5 million.

That subscriber total represents about 8.5% of MAU. At the time of Duolingo's IPO in 2021, the penetration rate sat just above 5%. It has climbed steadily since then, helped by the launch of new subscription tiers, the Family Plan, and a consistently richer product behind the paywall.

Subscriber growth relative to MAU, with penetration improving.

Share

At the Q4 2021 earnings call, von Ahn reflected on this penetration rate and shared his vision for how it could develop over time:

The 2025 Q3 report: shifting the balance

Taken together, the product mechanics, engagement loops, and LTV framework set the stage for understanding what happened next.

On the evening of November 5, Duolingo released its Q3 2025 report. The results looked strong with revenue, bookings, and EPS all coming in above expectations. DAU and MAU fell slightly short of forecasts, but not in any alarming way.

But the market's attention quickly shifted from the quarter's performance to management's guidance. By the close of trading on November 6, Duolingo's stock had plummeted 25%.

The answer was in a subtle but significant strategic shift and what that meant.

For the fourth quarter, Duolingo guided revenue, bookings, and adjusted EBITDA all roughly 5% below estimates. The bookings forecast indicated an annual growth of 22%, a marked deceleration from the consistent 30%+ growth rates the company had delivered over the past few years. Behind that softer outlook was what management called “a slight shift” in its short-term priorities.

As described earlier, Duolingo has long carefully balanced its key objectives: teaching better, driving engagement and user growth, and improving monetization. Experiments and new features were judged on their ability to move these metrics without hurting the others, a framework that had delivered strong performance across the board.

Now, that equilibrium is being nudged. The company says it is explicitly tilting more toward teaching efficacy and user growth, and relatively less toward near-term monetization. On the earnings call, CEO Luis von Ahn summarized the change:

This shift is already visible. DAU growth decelerated during the quarter, and the company guided for that trend to continue into Q4. However, CFO Skaruppa noted that broader platform retention “remains strong, with no real changes in that,” suggesting the slowdown is more about new users than existing learners dropping off.

At first glance, it can seem paradoxical that prioritizing user growth coincides with softer user metrics. But as described earlier, improvements in teaching take time to flow through the system. Although they are expected to temporarily slow growth, the hope is that they will ultimately increase engagement, user growth, and long-term LTV. Von Ahn explained the mechanism on the call:

The opportunity ahead

So why make this shift now? Von Ahn pointed to that “we are in a unique point in time”, a moment of structural change as a result of AI. And with its shifting balance, Duolingo is positioning itself for the next decade of learning:

Matthew Skaruppa framed the same shift through the lens of maximizing platform LTV:

Whether this strategic shift pays off remains to be seen. Luis von Ahn acknowledged that it “will be investing for a while,” and that the effects will unfold gradually. In other words, the kind of ambiguity that markets are rarely patient enough with in the short term.

Skaruppa provided more insight, underscoring that this is not a wholesale strategic pivot, but rather a slight reweighting of the same trade-offs Duolingo has successfully managed in recent years:

He also downplayed the near-term financial impact:

For now, the company is leaning into the long-term potential of its AI strategy, even if that means living with more uncertainty in the near term.

The bigger picture

Since its IPO in 2021, Duolingo has increased its revenue with a CAGR of 44%. Over the same period, the company moved from operating losses to full-year profitability in 2024. Being a platform company with very low customer acquisition costs and minimal incremental fixed costs, Duolingo benefits from significant operating leverage as its user base scales, which has been a major contributor to its expanding profitability.

While subscriptions remain the primary revenue driver, the remaining mix of advertising, in-app purchases, and the DET all show robust growth, collectively increasing revenue by 20% year-over-year.

Because stock-based compensation is a recurring part of operating expenses, free cash flow provides a clear picture of Duolingo's underlying profitability. Since Q3 2021 TTM, the company has expanded its free cash flow more than ~25 times, representing an increase in its FCF margin from 6% to 37%.

To get a perspective on the growth in the latter part of this period, FCF per share grew 49% year-over-year from Q3 2024 TTM to Q3 2025 TTM, from $5 to $7.4. As is evident, after the extreme growth rates, the pace has moderated. However, the growth rate still far exceeds what's needed to offset dilution and deliver meaningful per-share value creation.

The chart below shows this trajectory on a quarterly basis. One note on seasonality is that Duolingo runs highly popular campaigns around New Year's, which consistently boost Q1 results.

Duolingo's free cash flow per share, increasing since its IPO.

Share

With subscriptions and in-app purchases together accounting for roughly 85% of revenue, much of Duolingo's revenue flows through Apple's App Store and Google Play. For the nine months ended September 30, 2025, Apple processed 61.6% of total revenues, Google (Alphabet) 23.5%, and Stripe (primarily web-based subscriptions) 10.1%.

The first two of those app-store channels come with commission rates that can be as high as 30%. Duolingo's filings note that third-party payment processing fees from these distribution platforms make up the bulk of its cost of revenue. While that dependence has historically been a drag on margins, it could become a source of upside going forward.

Regulatory and legal pressure on mobile platforms has intensified globally, from Epic's litigation against Apple in the U.S. to enforcement of the EU's Digital Markets Act and similar rules in other markets, all of which challenge aspects of the traditional 30% app-store model. If, over time, these efforts lead to structurally lower effective fees on app-store billing or make it easier and more attractive for Duolingo to steer users toward lower-cost direct billing via Stripe and the web, much of that benefit would flow through to gross margin and profitability.

While the stock market's reaction to Duolingo's Q3 earnings might suggest the fundamentals are in free fall, the broader picture tells a different story. Growth is likely to slow somewhat as the company leans into longer-term investments, but at this stage, both interpretations are still on the table. The underlying product engine and engagement remain strong, but whether this strategic shift pays off will only become clear over the coming quarters and years.

Closing thoughts

Duolingo's ability to balance teaching quality, engagement, and monetization has taken it from an idea to a multibillion-dollar education company. Along the way, it wrapped that mission in streaks, characters, and a deeply data-driven product engine, improving with every lesson completed. Whether it's teaching languages, math, music, or chess, the goal is the same: make learning fun and worth coming back for.