Design and evaluate an ads ranking algorithm

## Ads ranking algorithm (sponsored content) You are designing an algorithm to rank ads in a feed/search results page. ### Requirements - Objective: maximize long-term platform value (e.g., revenue) while maintaining good user experience. - Constraints: low latency, advertisers have budgets/bids, user experience guardrails, and potential policy/fairness constraints. ### Questions 1. Describe a **ranking architecture** (candidate generation → scoring → final ranking) and what models you would use. 2. What would you predict (e.g., pCTR, pCVR, expected revenue), and how would you combine predictions with bids/budgets? 3. How would you handle common issues: position bias, calibration, cold start for new ads, and feedback loops? 4. Propose an **offline evaluation** plan (metrics + validation strategy) and an **online testing** plan. 5. List key monitoring metrics after launch and how you’d detect regressions or fraud/gaming.