ML System Design·Easy
Design a Search Ranking System
Asked at Google, Elastic, Algolia
Design a search ranking system for an e-commerce platform's product search. When a user types a query like "wireless noise cancelling headphones under $200," the system must return the most relevant products ranked by relevance, with considerations for business metrics like conversion rate.
Scale
- 50 million products in the catalog
- 10 million search queries per day
- Target latency: under 200ms for search results page
- Catalog updates: 100K product changes per hour (new products, price changes, inventory updates)
Design Requirements
- Design the overall search pipeline from query to ranked results.
- Explain your approach to query understanding — handling synonyms, misspellings, and intent.
- Design the ranking model — what features and what architecture?
- Describe how you measure search quality and run experiments.
- Discuss how to balance relevance with business objectives (promoting high-margin products).
Your Answer
Unlock AI-powered scoring, all questions, and progress tracking.