Design "Restaurants You May Know" Recommendation Algorithm
Overview
This question evaluates a data scientist's competency in recommender systems, product analytics, A/B experimentation, causal inference, feature engineering, and handling business constraints such as cold-start and attribution.
Meta
Jul 12, 2025, 6:59 PM
Data Scientist
Onsite
Analytics & Experimentation
20
0
Food-Delivery: "Restaurants You May Know" Recommendations
Context
You are working on a food-delivery app with a personalized home page. The team wants to add a new module called "Restaurants You May Know" to help users discover relevant restaurants they haven’t ordered from recently.
Assume standard clickstream and order events are available (impressions, clicks, add-to-cart, orders, order value), and the module links to the restaurant’s menu page.
Questions
- Why might users and the business need this feature?
- How would you design the recommendation algorithm (from candidate generation to ranking), including handling cold start and business constraints?
- How would you design the end-to-end user experience (placement, explainability, controls, feedback)?
- What experiment design and metrics would you use to evaluate the feature’s impact (include CTR, order conversion, retention, and an attribution window)?
Solution
Login required
Show