As the new adage goes, when the product is free, you are the product. The free language learning app Duolingo is interesting for three primary reasons. First, because it demonstrates how modern digital technology can offer new forms of instruction, specifically self-directed education, in a subject as common as language learning. Secondly, how in this age of free services can such technology be offered to the world on a sustainable business model. Lastly, the service must succeed in its goal of language instruction, and be engaging in order to attract a broad enough base of users to provide a business opportunity. There’s a “groundswell” in online education, commended Dan Weld, a professor of computer science at the University of Washington. “But a lot of it is prepackaged video clips and other things we tried a long time ago that didn’t go anywhere. We need more power and personalization” (Metz, 2012).
Duolingo has achieved growing popularity by creating a free, online tool, that is beautifully designed and fun to use. And I am curious on how they determined the design of their platform and the use of spaced repetition to achieve that success. But what is truly unique about Duolingo’s model, compared to the Big Data giants like Facebook and Google, is that this time a gigantic pool of users is exploited not to harvest their data, but to leverage their participation. Duolingo started out in 2011 as a project led by Carnegie Mellon computer science professor Luis von Ahn and one of his PhD students, Severin Hacker, two years before announcing it publicly (Product Habits, 2015). A few years earlier, in 2009, the language learning software company Rosetta Stone went public, raised $112.5 million on their first day of trading, and achieved over $209 million in annual revenue. However, von Ahn and Hacker determined that Rosetta’s high prices neglected a potential market of 800 million language learners, who for the most part were looking to improve their job prospects (Product Habits, 2015).
Duolingo’s early monetization plan was to sell the content that their users translated as they used the language learning program. von Ahn saw the opportunity based on his experience in developing reCAPTCHA, and spam deterrent that required users to type out a set of garbled letters to prove they were human. The company laid the early foundations for crowdsourcing, by also using end-users to verify words that computer scanners could not recognize. Google later purchased the company to exploit the same crowdsourcing approach to digitize books (TechCrunch, 2007).
Von Ahn had begun several years ago planning on how to translate the Web, and he thought his crowdsourcing approach, which he dubbed “human computation” (Simonite, 2012), made the most sense, since existing computer translations like Google Translate were ineffective. However, he needed a way to motivate people to participate in the project, and settled on the idea of teaching users a new language (Metz, 2012). Before they launched their beta, Duolingo had raised $3.3 million in a Series A led by Union Square Ventures (USV) and the celebrity tech-investor Ashton Kutcher. According to USV’s investment announcement, Duolingo’s pitch to investors was not so much about creating a free education tool, but more about building a sustainable way to generate human translations of online content on a large scale (Product Habits (2015). In 2012, Duolingo received many investments including a $45 million Series D round of investment led by Google Capital (Unicorn Economy, 2017). Duolingo has 95 staff members, of whom many were Google employees (Carpenter & Todd, 2014).
Therefore, when students practice by translating single sentences from one language to another in Duolingo, those sentences taken from sites such as Wikipedia. After multiple students translate the same sentence, software compares their results to compile a final version. These outputs are then accumulated and combined to create a translated version of an entire document. The results, explained von Ahn, are better than a computerized translation, and typically just under of professional quality (Simonite, 2012). By 2012, Duolingo had over 125,000 active users who translated 75 million sentences from Wikipedia and other online sources (Metz, 2012). In 2014, Forbes reported that that Duolingo was making money by selling crowdsourced translations to companies like CNN and BuzzFeed. von Ahn has additional ideas on how to further leverage the approach. For example, the work of students learning computer languages online on websites like Codecademy. “You could imagine something with programming, maybe finding bugs in software as part of a course,” said von Ahn. “We may try it” (Metz, 2012).
However, von Ahn claims crowdsourcing was never in Duolingo’s original business model, which was dropped in 2015, when the company turned its focus to serving a global audience of language learners (Coren, 2018). Duolingo has had to rely substantially on volunteer labor. Of the 100 or so employees, more than half work in engineering while just three people manage the volunteer community (Coren, 2018). Overall, only 6% of the company works on “product” and “research” (although a few language experts also consult with the company). This work is fed into the core of Duolingo’s language service, “The Incubator.” This language database allows hundreds of volunteers to write and translate thousands of phrases and sentences (Coren, 2018).
In 2013, Duolingo was named App of the Year by Apple. Today, Duolingo is now worth $700 million, and is the one of the world’s most popular language learning apps, among a field that includes the giant Rosetta Stone, to newcomers like Verbling, Colingo, Mermrise and FluentU. The app has about 300 million registered users across the world (Hartmans, Avery (2016). However, that valuation doesn’t seem justified by the company’s revenues, which were only revenues were only $1 million in 2016, growing to $13 million in 2017, and $40 million in 2018 (Lardinois, 2018). Given Google’s interest, it must be presumed that the company’s base of translators are of far greater value, likely to improve its own Google Translate software, then any financial profits it expects to share. And it is therein that we are reminded to not get caught up in the pretended value of a new digital learning solution, but remember how again we as consumers can become products.
References
Coren, Michael J. (2018, May 20). Duolingo’s crowdsourced language-learning model is letting some weird things slip through the cracks. Quartz. Retrieved from https://qz.com/1255133/duolingos-crowdsourced-language-learning-model-is-letting-some-weird-things-slip-through-the-cracks/
Conrad, Alex (2014, February 18). Language App Duolingo Raises $20M In Race To Teach English. Forbes. Retrieved from https://www.forbes.com/sites/alexkonrad/2014/02/18/language-learning-app-duolingo-raises-20m-in-race-to-teach-english/#187d91481cdf
Hartmans, Avery (2016, 24 March). Duolingo moving to East Liberty, plans to add employees.” The Pittsburgh Business Times. Retrieved from https://www.bizjournals.com/pittsburgh/news/2016/03/23/duolingo-moving-to-east-liberty-plans-to-add.html
Carpenter, M. & Todd, D. M. (2014, December 7). The Google effect: How has the tech giant changed Pittsburgh’s commerce and culture? Pittsburgh Post-Gazette. Retrieved from http://www.post-gazette.com/business/tech-news/2014/12/07/Google-effect-How-has-tech-giant-changed-Pittsburgh-s-commerce-and-culture/stories/201412040291
Lardinois, Frederic (2018, August 1). “Duolingo hires its first chief marketing officer as active user numbers stagnate but revenue grows”. TechCrunch. Retrieved from https://techcrunch.com/2018/08/01/duolingo-hires-its-first-chief-marketing-officer-as-active-user-numbers-stagnate/
Metz, Rachel (2012, June 25). Startup Has Language Learners Translating the Web. MIT Technology Review. Retrieved from https://www.technologyreview.com/s/428330/startup-has-language-learners-translating-the-web/
Product Habits (2015). How Duolingo Built a $700 Million Company Without Charging Users. Retrieved from https://producthabits.com/duolingo-built-700-million-company-without-charging-users/
Simonite, Tom (2012, November 29) The Cleverest Business Model in Online Education. MIT Technology Review. Retrieved from https://www.technologyreview.com/s/506656/the-cleverest-business-model-in-online-education
TechCrunch (2007). reCAPTCHA: Using Captchas To Digitize Books. Retrieved from https://techcrunch.com/2007/09/16/recaptcha-using-captchas-to-digitize-books/
Unicorn Economy (2017, March 28). How does DuoLingo make Money & Understanding DuoLingo Business Model. Retrieved https://unicornomy.com/how-does-duolingo-make-money-business-model/