What problem does it solve? AI-assisted product analysis often produces confident but wrong conclusions: vanity metrics, correlation treated as causation, overreaction to small samples, and unsegmented averages that hide the real story. This Skill structures the analysis process so it ends in a decision, not a dashboard. ## Core Features & Use Cases - Question-first analysis: Starts from the decision or question being informed, ties metrics to goals, and checks data trustworthiness before analyzing. - Bias and noise guardrails: Names and counters survivorship bias, selection bias, small-sample traps, and correlation-as-causation; enforces segmentation since averages lie. - Decision-ready output: Pairs quantitative findings with qualitative "why" and closes with an insight → impact → action recommendation. - Use Case: Ask why activation dropped last month. The Skill picks the right method (funnel, cohort, or segment), filters noise from signal, segments by device and acquisition source, and delivers an owned, measurable recommendation. ## Quick Start Analyze why our onboarding activation rate dropped last month and turn the findings into a specific recommendation.
