Subscription models live off habits: lessons from Duolingo’s retention success

Duolingo retention strategy Duolingo retention strategy

Subscription models live off habits. And hardly any app is as good at this as Duolingo! Let’s dive into what makes their strategy a success. 

Duolingo, a rapidly growing language-learning app with a market capitalization of almost $15 billion, owes a large part of its success to its streak feature. A streak is a count of consecutive days a user has completed at least one exercise. If a user misses a day, their streak resets to zero. This simple feature has made Duolingo one of the apps with the best retention, with an impressive 9 million people maintaining a streak of at least one year.

The company constantly optimizes this feature, running more than 600 tests over the last two years. Jackson Shuttleworth, Group PM of the Retention Team, shared key insights into their strategy in Lenny’s Podcast.

How does Duolingo ensure they work so well? 

Duolingo found a goal that their audience want to achieve and established rewards

Streaks don’t work for every digital product. They’re designed to build habits that users are proud of, that are difficult to achieve. But they only work if users have a clearly defined goal. 

For Duolingo, that’s: “I would like to do at least one language exercise every day”

Others have done this too: 

  • “I want to exercise for at least 17 minutes every day” (Freeletics)
  • “I would like to write a newsletter every week” (beehiiv)
  • “I want to find out about world events every day and be able to have my say” (news pages)

And this is established within the first minute of downloading the app – no lengthy signups, no overexplaining, just low friction and a focus on the goal: 

  • Hook: Choose a language right away.
  • Instant Reward: Complete a super short first lesson, then get a streak count, badges, and XP.
  • Personalization: Adapted the lesson path after just a few interactions.
Duolingo's habit loop
Duolingo retention habit loop

They lean into the psychology behind streaks

Streaks tap into two powerful psychological effects to build user habits:

Loss Aversion: This principle states that the pain of losing something is psychologically more powerful than the pleasure of gaining it. Users who have built a long streak are reluctant to lose it.

Sunk Costs: This effect describes our tendency to continue an endeavor once we’ve invested time, money, or effort into it. The longer a streak continues, the more a user feels they have invested, and the more valuable the streak becomes.

These two psychological drivers work together to increase a user’s perceived value of their streak with each passing day, which in turn boosts retention.

A balance between ambition and flexibility

While streaks motivate, there’s a danger in making them too easily accessible, which can devalue them. Conversely, losing a streak can cause retention to drop sharply. Duolingo addresses this by finding the right balance between ambition and flexibility.

  • Streak freezes: Allow users to take a day off without breaking their streak, preventing the streak from becoming so rigid that it’s worthless
  • Perfect Streaks: Rewarding users who manage a week without using a freeze, acknowledging consistent effort. 
  • Streak Repair: Users can restore a lost streak. This feature, once paid, is now free but requires users to complete additional exercises to earn the repair. This shift prioritises long-term user loyalty over quick monetisation

Streaks have now become something that language learners use to prove their commitment, to themselves and others, celebrating long streaks. 

Duolingo streak mechanism

A focus on key metrics and user commitment

Instead of just focusing on Daily Active Users (DAU), Duolingo optimizes for an unusual metric: the Current User Retention Rate (CURR). This metric measures the percentage of users active today who were also active in the past week. By focusing on the CURR since 2018, Duolingo has been able to increase its Daily Active Users by 4.5 times over four years.

The company also helps users commit to their goals. Users are regularly asked to set their next streak goal, with more ambitious goals offering greater rewards. The wording of a simple button from “next” to 

“I’ll commit” even led to an increase in retention values.

Duolingo growth model
Source

Remembering the importance of simplicity and timely reminders

Duolingo also understands the power of well-timed reminders. Initially, streaks were based on Experience Points (XP), but now a user only needs to do one exercise a day. This simplicity makes the feature easy to understand and encourages more frequent use.

Reminders are sent out 23.5 hours after a user’s last lesson, anticipating when they might have time for their next exercise. The reminders become increasingly urgent as the end of the day approaches, motivating users to complete their lesson.

These strategic optimizations, from psychological triggers to simple copy changes, are the key reasons why Duolingo’s streak feature is so effective and why the company continues to thrive.

An important final note: 

You can’t copy and paste ideas, but be inspired by them and ask how you could translate this feature. 

Streaks work so well on Duolingo because all users share the same goal: they want to learn a language.

With news apps, motivation is much more diverse: some users are looking for entertainment, others are news junkies who check the app three times a day, and some are only there to play Wordle. Before introducing streaks, you need to identify a goal that many users share — and one they are consciously aware of themselves.

Duolingo also relies heavily on gamification and humor. But this approach doesn’t suit every brand or audience. If you reward users with a comic-style animation and a streak for reading an article about the Middle East conflict, you’ll end up confusing them more than delighting them.

It’s therefore important to consider how the change in context can impact the success of a feature, to borrow ideas from other products intelligently. When looking to adopt a feature, ask yourself:

  • Why is this feature working in that product?
  • Why might this feature succeed or fail in our context, i.e. will it translate well?
  • What adaptations are necessary to make this feature succeed in our context?