How Duolingo's Experience Makes Language Learning Fun

As the world’s number one language learning app, Duolingo makes picking up a new language fun, free, and effective. 2020’s stay-at-home pandemic restrictions have seen its usage skyrocketing. March alone brought a 101% global increase in new users. Today, it boasts more than 300 million users who complete (on average) about seven billion exercises monthly.

As part of gamifying language education the brand leverages Duo, its charming owl icon, as part of its customer experience. Internally, it boasts a culture of constant testing and learning, and driving innovation by using AI and machine learning to continually improve the user experience. 

Customer data has helped shape features including tips and push notifications to help make language learning effective and keep users motivated. Data-driven insights help personalise the customer experience, including determining challenges within a practice session based on users’ success/fail rates.

The goal is to create a rewarding customer experience that makes language learning stick—and the effort is paying off. Duolingo’s latest research shows that its users perform as well on reading and listening tests as university students with four semesters of classes in the same language—but with half as many hours spent studying.

To find out more, R/GA chatted with Cem Kansu, director of product management at Duolingo, whose focus on customer experience helps drive growth, loyalty, and new revenue streams. An excerpt from our conversation:

Q> Cem, what problem is Duolingo solving with language learning?

Cem> The toughest part about learning a language is it's self-taught learning, and it's really hard to motivate yourself to keep studying. That's why a lot of the product decisions that we make as a brand is part of that, as well as how to make language learning more fun than any other tool out there.

Q> How do you humanise the Duolingo customer experience?

Cem> It begins with Duo, our owl mascot, who you see throughout your experience. He acts as your language learning coach and our kind of human voice. It takes the experience away from just being standard textbook language learning to one that is more gamified and humanised. He’ll motivate you to keep going if you're doing really well on lessons. And if you're struggling, he’ll say don't worry, keep going.

Q> How do notifications with Duo help re-engage users if their activity has dropped off?

Cem> The notification is a big lever for Duolingo because we are, at the end of the day, a daily habit application. If you can form a daily habit—and we see that happen roughly in three or four weeks—if you can stick around, build up to a streak and not miss a day, then you're basically going to stick around a lot longer. If you do a lesson on Duolingo today, you'll notice Duolingo will fire a notification right around the same time you completed your lesson the next day. This works really well for us.

Q> What role does gamification play in your CX?

Cem> We have a few mechanics that our users really love, and sometimes really obsess over. One of them is the Daily Streak. The best way to learn a language in any context is obviously doing daily practice, so we want to enforce that by showing our users how many consecutive days they've practiced the language on Duolingo. It really helps create good habits. A second one is leaderboards. It's a pretty simple competition mechanic—the more language you study, the more points you get, and that point is going to put you higher on a leaderboard. It motivates some of our users quite a bit to take the extra step to learn a bit more.  

Q> How are you using data, AI, and machine learning to engage and motivate users?

Cem>We have quite a few machine learning projects that are ongoing. Our push notifications have a lot of ML behind them. When you practice, it takes all of your user data and optimises what copy to use and when to send a push notification.  

The second one is when our users practice. So whenever you start a practice session that compiles what you've learned, the mistakes you made, the strength of the words—what we call a word strength model, which is if you made mistakes on a certain word, when you learned it, when you got it correct. It models how much you know that word and it caters the practice to that user.  

We're also working on ML projects that customise the language course to the user. The core content is the same, but the type of exercises shift so if users are really struggling, it makes it easier. If the users are basically crushing it, it makes it slightly harder.  

Q> Finally, what is your core experience as you look to scale Duolingo?

Cem> Continue making language learning fun; making Duolingo highly efficient to teach closer to fluency; and providing access to anybody that needs it, meaning we'll always have a great free, fun user experience.