Ground-up thoughts on building a consumer application
Key takeaways for founders from Duolingo's journey
Namaste Dear Reader
2025 has dawned upon us and I am back with a long-overdue edition of this newsletter.
One of the themes I have pondered over the last year as an early-stage investor is the implication of AI for consumer-facing applications. Funding has flown into apps that provide an ‘AI companion’ for everything under the sun: mental health, personal coaching, diet, dating, and many more.
Re-imagining AI-first applications is exciting. There are a lot of pieces on why these applications could expand the addressable market, and solve problems with more efficacy – however, I found a dearth of articles with fundamental ground-up information on what makes applications grow. This led me down a rabbit hole of chasing applications that have made it large successfully and exploring their strategies – this ultimately led me to Duolingo!
Duolingo, as most of us know, is the world’s leading language learning app and much more! To add context, the company is run-rating at $700 Mn+ in revenue and has grown its revenue at a CAGR of ~60% over the last 5 years.
Taking a step back, it’s incredible to see that a mobile application (that is not a gaming/ media company) drives home these numbers; especially when you factor in that a core of the revenue is via subscriptions (80%) and not advertising-based.
With over 800 million downloads, the company’s flagship app has organically become the world’s most popular way to learn languages and the top-grossing app in the Education category on both Google Play and the Apple App Store.
Other incredible stats that make Duolingo a company worth studying:
On Google, people search the term “Duolingo” nine times more often than “learn Spanish.”
Duolingo is growing its market – The company estimates that 80% of its users in the U.S. were not learning a language before using the app.
For Duolingo, on any given day, the number of returning users (i.e., those who have been away for more than a month) tends to be higher than the number of brand-new users who come to the platform.
Key takeaways from Duolingo’s journey:
First, build the world’s best free app in your category
Duolingo offers a significant portion of the app entirely for free – While they have added more paywalls/ advertising sections today, users can use Duolingo for as long as they like, and complete as many of our courses as they choose, all without paying anything. Learners who use Duolingo for free see an ad at the end of each lesson, whereas learners who purchase the premium subscription, enjoy an ad-free experience and access to additional features. Today Duolingo very much looks like a standard freemium offering, however, I don’t think its origin is appreciated enough.
Here is what Luis von Ahn, founder of Duolingo said in 2022 “As I reflect on our journey, I can confidently say that two of the best decisions we made when we first launched the Duolingo app were (1) giving it away for free and (2) making it fun. Rather than monetizing through a paywall or degrading the free user experience, we focused on creating an engaging free app that gamifies the language learning process. We believe this strategy allowed us to build a huge pool of learners who use and love our product, who tell their friends about it, and who we can convert to paid subscribers.”
Some other quotes:
“...it's why 90% of our user growth has been organic and why we feel so strongly that our free product must remain excellent …”
“The size of our user base is a big part of our data moat. Duolingo has more data on how people learn languages than any other entity in history, and we use that data to improve our effectiveness and engagement.”
While the freemium concept is now par for the course in most apps, the free portion often tends to be a meagre imitation of the entire application. In contrast, Duolingo’s free version would still rate very highly vs the paid offerings of its competitors.
A point to note here is that Duolingo started experimenting with a paid version of its app after perfecting it for 2 years; while it started focusing on subscriptions aggressively only from 2017 – 6 years after its launch!
First – Build the best App in your category!
In today’s ultra-competitive world, the first problem you are trying to solve is to get to certain mass of users onto the platform. Unless you have a bucketload of cash to keep advertising, the only way the app can grow is if it is:
Free + Effective + Engaging
More on the holy trinity below
So when is the right time to charge a customer?
In my view, when the app seems too good to be free. This can be as soon as after a couple of months to a couple of years of development as well.
An under-appreciated facet here is that a business that monetizes too soon leaves the door open for a competitor to enter!
Closer to home, Indiamart (India’s largest digital B2B marketplaces) continues to allow buyers free access to its platform whilst offering a base-level package that includes a listing to all sellers for free.
Globally, ChatGpt introduced the world to AI and yet to continues to offer a base offering entirely for free to drive more users and data onto its platform.
I am not advocating for startups to push monetization into the future for a similar period – most startups don’t have the luxury of the quantum of VC money afforded to the company. What I am proposing is that founders initially focus on building the best-in-class product and then look to capture a portion of the value offered.
No dichotomy exists between user growth and monetization
One of the common quips I hear from founders is that a startup can only focus on growth or monetization – While in the short-term this may be true over the long-term monetization almost always flows from user activity & growth, especially in a consumer app.
In a sense, this point is a corollary to the previous one – startups need to think deeply about whether they add enough value to their free version to drive growth. This is especially relevant for AI-first applications where data is the oil!
Too often, founders chase monetization aggressively when growth starts to stall. This is akin to killing the golden goose that lays the eggs. While there is some merit in pursuing this strategy once a company corners a large chunk of its addressable market, this almost always does not apply to a startup.
Duolingo explains this via its flywheels:
Learning flywheel: The greater the scale of the learner base, the more insights can be drawn to improve both engagement and efficacy. Higher engagement leads to better teaching, which results in retention and referrals – leading to more users.
Investment flywheel: The more learners use Duolingo and convert into paid subscribers, the more the company can invest in creating an even more delightful, engaging, and effective learning experience
The way I’d visualize this for a founder is this: Active Users in the app are its engine, while the efficacy+ engagement is the axle that moves the wheels of monetization.
Prioritize being a technology company first, then a marketing company before being an advertising one
It is often very easy for consumer applications to morph into advertising companies – in a ZIRP era, this strategy worked well – where cheap venture money fuelled customer acquisitions. However, this almost always never works in the long run. Funding during COVID created several such consumer tech companies that focused on advertising without really building technology that delivered value. Consumer behaviour reverted sharply to offline activities and we have seen the shutdown of several such apps – especially in the edtech sector.
Questions for startup founders to answer:
Does your tech deliver value for the customer?
What is your product philosophy?
What is your strength/ drawback vs a similar offline offering – Does your application leverage this?
Duolingo defines itself as a proud technology and product-driven company—70%+ of its employees work directly on improving products. In 2021, the company famously claimed to run about 500 A/B tests every quarter – we can safely assume that number has only gone up significantly gone up over the last 4 years.
Duolingo’s R&D spend as % of revenue has hovered around the 30 to 40% mark despite 10x growth in revenue:
Being a consumer offering, it is essential to build your brand – marketing and brand salience can’t be ignored. However marketing does not mean buying users – Between 2011 to 2019, Duolingo spent less than $15Mn in aggregate on external marketing. The company rather focused on building its iconic mascot, Duo the Owl and invested in improving its design and storytelling elements. More recently, the company has used TikTok in the US to good effect to retain its virality and mind-share: One of the main things people share are weird sentences that they see on Duolingo.
Focus on retention – Not Monthly, Weekly but Daily.
No article on Duolingo is complete without talking about gamification streaks and their legendary retention strategies. The company understood early that the hardest part of learning something new is staying motivated, so they decided to make Duolingo as fun as possible.
The company uses gamification mechanics like a streak that counts how many days in a row you've practiced on Duolingo. Back in 2021, over 50% of Duolingo’s daily active users had a streak longer than 7 days, and around 1 million learners have an active streak longer than 365, meaning they have used Duolingo daily for at least a year or possibly longer.
Duolingo’s greatest gift to the app world is its relentless focus on DAU (daily active users) – A user that uses your app daily is far more likely to engage over a longer and convert into a paid subscriber. Duolingo discovered that the single most important lever for improving DAUs over time is the Current User Retention Rate (CURR). A point to note here is that Current Users are learners who are active today, who were also active in the past week – this specifically excludes new users and also excludes users who haven’t used the app in the past week. Essentially this narrows down the focus of the team to potential power users of the app.
Measure what matters!
The final takeaway I’d like to leave with every startup founder is the operating metrics that should be tracked.
There are 3 groups of metrics that every business needs to define:
Efficacy metrics
Growth metrics
Monetization metrics
Efficacy metrics are usually the hardest to define given that it becomes difficult to demonstrate tangible outcomes for the users – However, this is precisely why apps fail or businesses fail. They don’t measure the value delivered.
Duolingo claims you can learn as much as 5 semesters of university instruction in just 5 sections of Duolingo.
However, it is also essential for the app to demonstrate efficacy to its users – Duolingo uses brief quizzes and levels at every interval to reinforce progress for users.
For all its buzziness, AI-companion apps will need to deliver AND demonstrate value to customers for a chance to succeed!
To summarize, here are my key takeaways for startup founders from Duolingo’s journey:
- Build the world’s best free app
- User growth and Monetization are two sides of the same coin
- Prioritize being a tech company
- Focus on retaining your daily power users
- Measure what matters
If you are a founder building a consumer AI application, I’d love to hear from you. Reach out to me on LinkedIn or Twitter.
I would love your (anonymised) feedback! Do share your comments & questions below or write to me @SaneDhruv on Twitter
Great read as always, urging you to write more! How do you think AI-first consumer apps should balance the resource intensity of developing groundbreaking tech with the urgency to demonstrate value early? What would you like to see being prioritize - emphasis on gamification? daily engagement?