Interactive Study Bites
Putting the first five multiple-choice questions of a set directly in Quizlet’s mobile home feed.
Situation
Students were coming to Quizlet but not studying. The mobile home feed was a place to browse sets and recommendations, but getting from opening the app to starting a session still took intent and a tap or two.
Conflict
Learn mode is Quizlet’s highest-engagement, highest-retention feature, and past data showed users who enter it are significantly more likely to stay. But entering took a commitment. Students had to decide to study, navigate in, and start. New users did not know what Learn felt like, so they scrolled past it.
Approach
Instead of better invitations, we tried removing the door. Interactive Study Bites (ISBs) put the first 5 multiple-choice questions of a set inline in the scroll, with no decision required. To keep it clean at a glance, eligible sets were English-language, without diagrams, with questions under 80 characters and answers under 32. We ran a 50/50 A/B test.
Turning point
Only 22% of treatment users ever saw an ISB because of feed ranking and set eligibility, which diluted our headline metrics. Among those who did see one, 24% tapped in, second only to Recents and Progress cards in the whole feed. Of those who started, 71% finished all five questions, and of those, 41% continued into full-screen Learn.
The reframe: this was not a weak feature with modest lift. It was a highly effective feature most people never saw. The constraint was reach, not quality.
Resolution
The test cleared the do-no-harm bar with room to spare: engaged learning at 7 days up 0.8%, subscriptions neutral, retention steady. For new users the effect was much sharper, with 7-day engaged learning up 11.6%, a sign ISBs help most where there is no study habit yet. We rolled ISBs out to all users and treated the result as a floor.
Impact
Of 503,182 users who saw an ISB, 34,573 (about 7%) entered full-screen Learn directly through it. For new users, ISBs outperformed every other educational recommendation in the feed. Next: broader feed ranking, personalizing questions with progress data, and more question types in 2026.
What I’d do differently
I would address set eligibility and feed ranking in the same sprint as the launch. The constraints that capped reach at 22% were knowable upfront, and fixing them in parallel would have given a cleaner read on true impact and faster sample sizes for new users.