How would you launch recommendations when both users and items have little interaction history?
Make your call first. Use the published answer and rubric to examine your reasoning when you are ready.
Case briefRead the inputs, then make your call
A learning app needs useful recommendations for new users and newly published lessons. The randomized pilot is fictional.
| Assignment | 500 new users per arm; keep original assignments for 14-day measurement. |
|---|---|
| First useful lesson | Baseline: 120 users. Candidate: 170 users. |
| Return and useful completion within 14 days | Baseline: 90 users. Candidate: 80 users. |
| Constraints | Little interaction history; onboarding preferences are optional. |
Your task. Design the recommendation and exploration approach. Interpret both outcomes and choose the next rollout decision.
0 words