What I Learned Reading Duolingo's Annual and Quarterly SEC Filings
Learning About Business, AI and Product Building Through SEC Docs
I’ve become increasingly curious about different businesses, so I’ve been spending more time reading annual and quarterly filings with the SEC.
Part of this is driven by general curiosity about how companies operate, particularly if they have influenced my life (for example as a customer).
Part of it is because I want to strengthen my understanding of finance and other business fundamentals.
And part of it is for inspiration as a product manager and builder working at the intersection of contracts, business and technology (mix of Generative AI, traditional machine learning, and deterministic software).
I started with Duolingo — in part because I’m a fan of the products and company, and in part because there are many parallels between contracts and language learning.
With that, here are some things I’ve learned reading Duolingo’s recent 10-Q and 10-K, and some related open questions that I have:
1. AI Use – Duolingo uses AI in a variety of ways, including personalizing lessons (e.g. predicting the likelihood that a user will get a question correct to help drive the optimal level of difficulty) and using generative AI capabilities in its Video Call product feature. Other examples include significantly accelerating content / lesson creation by leveraging LLM capabilities. Separately but relatedly, they were one of the first companies I saw or heard talking about becoming AI-first.
2. Large Data Moat – They’ve collected a massive amount of data corresponding to over 130 million active users. This in turn helps improve the user experience by personalizing lessons and otherwise improving the product, per the above.
3. Financial Metrics – They’ve experienced significant revenue and profitability growth driven by an engaging product experience. It’s interesting that revenue growth for both the quarter and the year is ~41% YoY despite some possible seasonality in the business. It’s also interesting that their DAU growth is 40% YoY — unclear if this is a coincidence or if there’s a more direct connection between these metrics.
4. User acquisition – This is primarily through word of mouth (driven by an engaging product experience that uses gamification and bite-sized lessons) and brand marketing (e.g. they’ve had some pretty epic social media marketing). However, there is an increasingly complementary use of paid user acquisition as well, for example, in high-opportunity markets experiencing limited organic growth.
5. Mobile as a distribution strategy – Early on, they bet that mobile would be a way of getting to as many learners as possible (target users of the product) due to the ubiquity of smartphones. This has paid off significantly as they continue to enjoy significant growth in active users and paid subscriptions.
6. Company culture – They champion their culture as being distinct from a more typical Silicon Valley culture. I’m not 100% sure what this means, but it could refer to being more disciplined about aggressive investment to drive growth (e.g., not hiring aggressively during the COVID pandemic/lockdown), living in a more affordable geography, and/or being more ideologically diverse.
Open questions that I have:
1. New hardware implications on Design & UX – There’s some interesting AI-enabled hardware innovation on the horizon, including Meta’s VR glasses as well as the OpenAI/Jony Ive collaboration. Is there any plan to start designing for that medium? If I had to guess, it would take years of change management and user adoption to shift from mobile (similar to how it took 10+ years to shift from desktop computers to mobile — though a remarkable new user experience could accelerate this).
2. Scalability and Profitability Growth – How much of the profitability growth is driven by one-time or non-recurring events? For example, it looks like costs may have decreased as AI automated some processes, but were any of these one-time decreases? Also, is there a view that features driven by generative AI will drive cost efficiencies as foundation models continue to benefit from performance increases and economies of scale? Finally, given the increasing need for paid user acquisition, is there a concern that profitability will suffer in the future as a result of increased customer acquisition costs?
3. Geographic/cultural advantages – What are specific examples of how the Pittsburgh geography/non-Silicon Valley culture manifests, and what are the pros and cons relative to Silicon Valley or another large city/tech hub?
4. Financial Metrics Breakdown by Region – What do revenue and profitability growth look like by region?