5 product lessons we learned from building Friend Streak

Jackson Shuttleworth20 Sep 2024

Problem

A product team launches a social feature whose value grows with the number of connections per user. It must choose how many connections to allow and which part of the adoption path to improve first. Testing each higher cap (6, 7, 8 slots) is costly, and a bigger cap needs more backend and screen space. At Duolingo, 57% of users have at least one in-app friend, so the team wanted a shared daily streak with up to 5 friends.

Example from the article: Duolingo compared retention for users with 1 to 5 shared streaks, saw that the few users with many active friends already retained well, and postponed raising the cap above 5.

Solution in brief

Build a throwaway local prototype to test the idea cheaply. Design the UI for several friends, but launch with a cap of 5. Before testing larger caps, analyze retention and group size across the existing slots. Then split the adoption path into steps and work on the step that blocks the most users.

Questions this case answers
  • How to decide whether to expand a capacity limit of a feature when only a small share of users ever reaches that limit
  • How to pick the funnel step that gives the largest gain in exposed users for a feature whose value comes from user-to-user connections
  • How to decide whether to raise the cap on a social feature (friends per user) before paying for experiments?
  • How to find which step of a feature's adoption funnel to optimize first?

Tools and methods

  • Cohort and retention analysis — retention and size of groups by number of shared streaks, to forecast the value of a higher cap
  • Funnel analysis — a six-step funnel from eligibility to five daily streaks, to find the biggest hurdle
  • Local-only prototype — an internal version with no backend, for dogfooding
Continue reading — it's free

Google or a one-time email code. No password.