Evaluate Metrics for Restaurant-Feature Impact and Engagement Trade-offs

Quick Overview

Evaluates product analytics for restaurant recommendations that may cannibalize new-friend suggestions. Strong answers track user and restaurant-side value, interpret flat engagement with a 2% friend-add decline, and propose targeting, placement, or ranking mitigations.

Evaluate Metrics for Restaurant-Feature Impact and Engagement Trade-offs

Company: Meta

Role: Data Scientist

Category: Analytics & Experimentation

Difficulty: medium

Interview Round: Onsite

##### Scenario Social app launches a restaurant-recommendation feed that may compete with existing new-friend suggestions; product team wants to understand engagement trade-offs across features and page types. ##### Question Which metrics would you track to evaluate the restaurant-recommendation feature from both the user side and the restaurant side? Overall engagement stays flat but the new-friend add rate falls by 2 %. How would you interpret this change? What product or targeting changes would you propose to mitigate any cannibalization between restaurant recommendations and new-friend suggestions? Activity on group pages is higher than on celebrity and family pages. Generate hypotheses to explain this difference and outline how you would validate them. ##### Hints Tie metrics to user value, surface cannibalization hypotheses, segment by audience, propose experiments or ranking changes to balance resource allocation.

Overview: Evaluates product analytics for restaurant recommendations that may cannibalize new-friend suggestions. Strong answers track user and restaurant-side value, interpret flat engagement with a 2% friend-add decline, and propose targeting, placement, or ranking mitigations.

|Home/Analytics & Experimentation/Meta
Meta logo
Meta
Jul 12, 2025
mediumData ScientistOnsiteAnalytics & Experimentation
86
0

Evaluate Metrics for Restaurant-Feature Impact and Engagement Trade-offs

A large social app launches a restaurant-recommendation feed that may compete with existing new-friend suggestions across page types such as group pages, celebrity pages, and family pages. The product team wants to understand engagement trade-offs and avoid cannibalization.

Constraints & Assumptions

  • Evaluate both user-side and restaurant-side value.
  • Separate total engagement from shifts between features.
  • Include guardrails for social graph growth and content quality.
  • Treat flat overall engagement with a friend-add decline as a meaningful trade-off to interpret.

Clarifying Questions to Ask Guidance

  • Where is the restaurant feature shown relative to new-friend suggestions?
  • What action is considered success for restaurant recommendations?
  • Are restaurants businesses, pages, or external destinations?
  • Is the product goal engagement, local discovery, revenue, or retention?

Part 1 - Metrics

Which metrics would you track to evaluate the restaurant-recommendation feature from both the user side and the restaurant side?

What This Part Should Cover Guidance

  • User metrics such as reach, impressions, CTR, saves, shares, directions, reservations, dwell, repeat use, retention, hides, and satisfaction.
  • Restaurant-side metrics such as page visits, follows, calls, reservations, orders, reviews, and merchant distribution.
  • Guardrails for new-friend suggestions, core engagement, spam, quality, and cannibalization.

Part 2 - Interpretation

Overall engagement is flat, but the new-friend add rate falls by 2%. How would you interpret this change?

What This Part Should Cover Guidance

  • Cannibalization versus harmless substitution.
  • Segment and page-type analysis.
  • Long-term impact on social graph growth, retention, and user value.
  • Statistical and practical significance.

Part 3 - Mitigation

What product or targeting changes would you propose to mitigate cannibalization?

What This Part Should Cover Guidance

  • Placement, ranking, targeting, frequency caps, page-type-specific treatment, personalization, multi-objective ranking, and holdouts.
  • Experiments to validate mitigations.

What a Strong Answer Covers Guidance

A strong answer tracks incremental value and substitution, interprets friend-add decline in business context, and proposes targeted product changes that preserve restaurant discovery without damaging social growth.

Follow-up Questions Guidance

  • What if restaurant clicks are high but downstream actions are low?
  • How would you detect page-type-specific cannibalization?
  • What if the 2% friend-add drop affects only new users?
Loading comments...