Root-Cause Analysis for an 8% Decline in Google Search Usage

You are a product manager working on Google Search. A dashboard shows that Search is 8% below where it should be, and you are asked what you would do about it. Explain how you would work out what caused the decline and how you would respond once the cause is known. The prompt does not say which Search metric fell, over what period, or how quickly, so settle those points before you build your list of causes. ```hint Read the shape of the drop Find out exactly what is 8% lower and how it fell over time. A step change on a single day and a slow slide over several weeks point to different families of causes, so the shape tells you where to look first. ``` ### Clarifying Questions - Which metric is 8% lower: total queries, daily searchers, queries per searcher or search sessions, and is the comparison day over day, week over week or year over year? - Did the decline appear abruptly on one date or build up gradually, and has any of it recovered since? - Is the drop concentrated in one surface (desktop web, mobile web, the Google app, the Chrome address bar, voice entry points), one country or language, or one user group, or is it spread evenly? - Did anything change around the start of the decline: a launch or experiment, a logging or bot-filtering update, or a change to where Google is the default search engine? ### What a Strong Answer Covers - Validating the number: ruling out instrumentation, pipeline, bot-filtering and metric-definition changes before treating the drop as a change in user behavior. - Separating real change from normal variation: seasonality, holidays, day-of-week mix and unusual events in the comparison period. - Decomposing and segmenting the metric (searchers versus queries per searcher; surface, entry point, platform, geography, language, query type and cohort) to localize the decline. - Ordering internal and external hypotheses by the shape and location of the drop: releases, ranking or results-page changes, latency and outages, distribution deals, competitors including AI assistants, news cycles, regulation and privacy sentiment. - Confirming the root cause with evidence and turning it into action: rollback or fix, owners, guardrail checks and monitoring that catches the next drop sooner. ### Follow-up Questions - Segmentation shows the decline is concentrated among younger users and built up over several weeks, with no release lining up with it. Which hypotheses move to the top, and how would you test them? - Queries are down 8% but daily searchers are flat. What does that tell you, and could the drop be a sign that Search got better rather than worse? - How would you set up alerting so that a decline like this is detected and localized within a day next time?