Metrics for Search and Recommendations

What metrics would you track to measure the success of an algorithmic search or recommendation feature?

Discuss how you would evaluate both search and recommendation experiences, distinguishing user engagement from genuine relevance and business impact. Explain which metrics you would prioritize, how you would calculate them, and which guardrails would prevent optimization toward low-quality outcomes.

1. Define a primary success metric and supporting metrics.
2. Describe the funnel from impression to meaningful user outcome.
3. Explain how you would segment results and validate causality through experimentation.
4. Identify potential failure modes, trade-offs, and guardrail metrics.

Available data includes impression logs, queries, ranked results, clicks, saves, purchases or downstream actions, session events, experiment assignments, latency, and user or content attributes.