Churn Analysis PDCA Framework
A structured human-AI collaboration process for monthly SaaS customer churn analysis. Cycle frequency: Monthly (one full analysis cycle per calendar month).
How to Use This Skill
Work through phases in order. Each phase has a STOP condition before proceeding.
PLAN — Define the analysis scope and confirm data sources before touching any data.
See references/phase-prompts.md → PLAN phase.
DO — AI runs data wrangling and surfaces patterns. Human writes interpretation and recommendations.
See references/phase-prompts.md → DO phase.
CHECK — Verify analysis integrity and recommendation quality before the leadership presentation.
See references/phase-prompts.md → CHECK phase.
ACT — Retrospect and propose refinements to this skill.
See references/phase-prompts.md → ACT phase.
See references/working-agreements.md for current version and process discipline rules.
STOP Triggers — Intervene Immediately
- AI writes or implies any explanation for why a churn pattern exists
- AI drafts recommendations or action items without human-provided interpretation
- AI presents pattern findings as conclusions rather than observations
- AI skips the data quality check and proceeds directly to analysis
- AI generalizes from a single cohort to a population-wide conclusion
Human Ownership — Non-Negotiable
- Interpreting what churn patterns mean (requires business context not in the data)
- All recommendations to leadership — AI formats, human authors
- Deciding which patterns are significant vs. statistical noise
- Any narrative connecting data to business events (product launches, market changes, etc.)
- The framing and tone of the leadership presentation