Evaluating the Facebook ‘Memory’ feature
Overview
This question evaluates a data scientist's proficiency in product analytics, experimentation design, and impact measurement by probing framework design, metric selection (retention, engagement depth, share rate, cannibalization, safety/quality guardrails) and CTR benchmarking.
Meta
Jul 12, 2025, 6:59 PM
Data Scientist
Technical Screen
Analytics & Experimentation
68
0
Evaluating the Facebook Memories Feature
Context
You are a Data Scientist asked to assess whether the Memories feature (which resurfaces users' past posts) delivers real user value without simply reshuffling attention from other content. You must design an evaluation plan, set acceptable CTR expectations, choose appropriate impact metrics, and avoid over‑resurfacing the same content.
Questions
- Framework: Outline how you would evaluate Memories performance. Consider retention, depth of interaction, share rate, cannibalization, and safety/quality guardrails.
- CTR Benchmark: Is a 20% click‑through rate acceptable for Memories? Explain how you would benchmark it against internal and external norms, impression volume, and user intent.
- Time‑Spent Metric: Is total time‑spent a good metric for Memories’ impact? Why or why not, and what should you use instead?
- Dedupe/Frequency: After resurfacing a 2019 post in 2020, how would you prevent it appearing again in 2021? Propose a deduping and cool‑down strategy.
Solution
Login required
Show