Behind the product: Duolingo streaks

The more that we can make the feature easily comprehensible to users, the more retentive it is. We've run a number of experiments to do this.

We really resist the urge to do the big V1. Rather than design the big complex feature for V1, just do the simplest encapsulation of what that feature can be, see if it has legs, and then just add to it iteratively over time.

Once you get to seven days, loss aversion kicks in, and you retain. Going from a one to a two-day streak, huge jump in retention, two to three day streak, slightly less but still huge and it's up until day seven.

Give more flexibility when a user is starting their streak. Eventually, once people get on long streaks, you don't want to give them as much flexibility.

What we realized when we ran this experiment is DAUs moved not one bit. The users that we were capturing were the least engaged users imaginable.

Duolingo tested adding an opt-out button to their streak goal feature, where users commit to maintaining daily learning streaks of specific lengths.

It was a huge win to let them do that. The learning here was that this intentionality of saying, no, I want it was where we were getting so much of the engagement from this feature.

I'd say test everything, we've run in the last four years over 600 experiments on the streaks, so every other day.

We send two notifications related to your core streak each day, the first is a practice reminder, we send it 23 and a half hours after you practice the day before.

Because people care about their streak, that notification reminding them, hey, come back and... People see this by and large as a positive notification and not a negative notification.

Give more flexibility when a user is starting their streak. Eventually, once people get on long streaks, you don't want to give them as much flexibility.