product-ops-analysislisted

# Product Ops Analysis ## Goal Turn product and operations data into reviewable product decisions without overclaiming data quality, exposing sensitive data, or treating exploratory analysis as an approved metric standard. ## Workflow 1. Define the decision the analysis must support, not just the chart to produce. 2. Identify the data source type: CSV or spreadsheet export, analytics tool, BI dashboard, warehouse, database, notebook, support tickets, CRM, ads platform, or manually supplied summary. 3. Load existing metric definitions and tracking taxonomy when available. If none is found, label metrics and events as proposed. 4. Classify data sensitivity and choose the safest access path: local exported files before live systems, read-only database or warehouse credentials before write-capable credentials, aggregated or sampled data before row-level customer data, and synthetic data for demos and public examples. 5. Use `skills/tool-vetting/SKILL.md` before connecting to analytics, warehouse, CRM, support, ads, or automation tools. 6. Record query or transformation assumptions, date ranges, filters, exclusions, timezone, identity grain, and cohort definitions. 7. Separate observed data from interpretation and recommendation. 8. Include uncertainty: missing events, sampling, bot/internal traffic, attribution gaps, delayed ingestion, experiment imbalance, small sample size, or changed instrumentation. 9. Produce decision-ready outputs: metric summary, segment/cohort differe