cyberneticlibrary

Analyze metrics into product decisions

data-analysis-standardskillsetup L1327
mohitagw15856/pm-claude-skills
What it does

Convert raw numbers into product decisions with root-cause analysis

Best for

Explaining unexpected metric movements before escalating to leadership; answers 'what changed, why, so what, now what' for any metric problem.

Inputs
  • · Metric or question being investigated
  • · Time period and current vs. baseline values
  • · Available segments (platform, cohort, channel, geography)
Outputs
  • · Root-cause analysis with evidence (not just hypothesis)
  • · Confidence level and recommended actions with owners and timelines
Preconditions
  • · Question clearly stated (not vague like 'why is revenue down')
  • · Baseline or benchmark available for comparison
  • · Access to segmented data (can't analyze by platform if data is aggregated)
Failure modes
  • · Confuses correlation with causation (names pattern but not cause)
  • · Over-confident conclusion without stating what data is missing
  • · Recommends action without owner or timeline
  • · Analyzes aggregated data — masks important segment-level shifts
Trust signals
  • · Four-question framework required: what changed, why, so what, now what
  • · Root-cause hypothesis with evidence (data point, not speculation)
  • · Confidence level stated and justified
  • · Explicitly names what the data cannot tell you
  • · Recommended action includes owner and timeline