Tired of the rote memorization of Latin? Noli timere — Duo the green owl and friendly amici are here to help. With a continuous stream of check marks, dings, and other bouncing buttons, you’ll never feel so affirmed in your life as the moment you successfully discern between dative and ablative nouns.
Founded in 2011, Duolingo is best known for gamifying the early language acquisition process, and has successfully scaled into a public business with an IPO in 2021. Beyond digitizing and scaling the language-education process Duolingo brought a new business model to the industry — fremium — which provides a level of scale and volume of users and contributors a base from which to develop unique capabilities that allowed them to win during the Covid-19 pandemic.
The Duolingo Business Model
Duolingo’s core product is a fremium interactive web platform that makes learning new languages fun. Rather than memorizing lists of words or conjugations, Duolingo’s lessons are entirely interactive. To encourage the consistent progress necessary for language retention, the platform awards users points over time for continuous streaks and performance on various lessons and tests. By signing up for a subscription (Duolingo Plus), users unlock additional features (streak repair, unlimited skill tests, and mistake reviews) and avoid ads. In comparison, Rosetta Stone and Babbel focus on subscription-based softwares and larger enterprise customers like schools and businesses (see Figure 1)
Validation of this approach to language learning is in the size of their user base, which reached a size of almost 10MM daily active users by the end of 2021 (Company 10K Report), including the likes of Bill Gates, Jack Dorsey, and Khloe Kardashian. As quoted in Forbes 2019, founder and MacArthur genius Luis von Ahn reflects: The moment I felt proudest was when I realized, “Wow, the richest man in the world is using the same system as the lowest people on the economic scale,” That is for me really special and pretty big.
The company relies on several network effects from this large user base in order to create value. For example, for the first 10 years of its life, the platform relied upon contributor communities to help put together language lessons for less common languages. These contributors worked for free, likely motivated by the ability to have impact at scale and also create more speakers of languages that they cared about. (The program ended in 2021). This allows Duolingo to offer “little-spoken tongues like Hawaiian, Navajo and Gaelic and the fictional language High Valyrian (from the hit HBO series Game of Thrones)” — enabling them to outcompete competitor offerings through lower costs (Forbes).
But perhaps the most important aspect of this scale is Duolingo’s ability to develop a unique AI competence in the language-learning industry. As described in VentureBeat, the company has found many ways to integrate predictive models into improving their core language learning product. For example, Duolingo has a model called Birdbrain that predicts a users likelihood of success for any particular exercise or test. Leveraging this model, the company is able to dynamically generate lessons that are at the best level of difficulty for learning — not too hard, but not too easy. The company has other models that optimize timing for repeating words, identifying the reasons a users got something wrong, or handling free-form text inputs and assessing their correctness. The key point here is that these kind of models benefit tremendously from the user volume that Duolingo’s freemium model enables.
Duolingo and the Pandemic
Given its digitally delivered product, Duolingo’s core product actually benefitted from the pandemic in the sense that more time at home led to a temporary increase in the company’s core metrics as people tried to make productive use of their time (Company 10K Report).
But the most significant benefit to Duolingo came from an adjacent product: the Duolingo English Test (DET). Started in 2016, the test provides an alternative to traditional language tests like the TOEFL or IELTS. However, unlike those test, DET is taken at-will, at-home, and for a fraction of the cost and time (See Figure 2). The key insight from Duolingo is that they could develop human-in-loop-algorithms that monitor test takers via their home computer’s camera and ensure users are not cheating during the test. DET revenues increased due to the pandemic because Covid-19 forced universities and other test-requiring institutions to accept the DET as a legitimate evaluation of english-speaking ability due to DET’s remote nature. Prior to 2020, universities had been slow to accept the test; but at this point in 2022, the test is accepted by leading institutions like MIT, Yale, and Duke. Duolingo has expressed confidence that these decisions to accept the test are not temporary: “we believe universities are unlikely to stop accepting the test when the pandemic ends” (Company 10K).
In summary, Duolingo is a winner of the pandemic because it is a digital-native, data-native company in an increasingly digital- and data- driven world. The pandemic assisted Duolingo in driving acceptance for their English Test product, but more generally Duolingo’s success comes from the unique design of its business model that enables its services to continually improve and expand. If the company is able to solve later-stage language acquisition, they’ll truly be in Terra Incognita.