Google Search Success Metrics: North-Star, Supporting Signals and Guardrails

Explain how you would judge whether Google Search is succeeding. Define what success means for the product, propose the primary metric you would use, and describe the supporting and guardrail metrics you would track alongside it. ```hint What does a satisfied searcher do? A search engine that answers well can produce fewer clicks and shorter visits. Think about what a searcher who got what they needed actually does before you pick a primary metric. ``` ### Clarifying Questions - Which surface is in scope: web search on desktop and mobile, the Google app, or also voice, image and AI-generated answer features? - Is the metric for an overall health review of Search, or for judging individual ranking and feature launches? - Should advertising revenue count toward success, or be treated as a constraint? - Should we measure globally, or for a particular market, language or user segment? ### What a Strong Answer Covers - The purpose of Search stated as a user goal, and a definition of a successful search that also covers needs answered directly on the results page. - A primary metric that reflects whether people found what they needed, with an honest look at how it could be gamed or misread. - Supporting metrics along the search journey: forming the query, the quality of results, speed, and whether people come back. - Guardrails on ads, latency, spam and misinformation, and traffic sent to websites. - How the metrics are validated, for example with human relevance ratings and controlled experiments. ### Follow-up Questions - After AI-generated answers launch, clicks to websites fall while your primary metric rises. Is that a success? - Queries per user rose last month. Is that good news, and how would you find out? - How would you choose the success metric for a single ranking change before it ships?