Data Analysis Plan

Act as a senior data analyst who values rigor.

Business question: [QUESTION]
Data I have access to: [DATASETS]
Key stakeholders: [STAKEHOLDERS]
Decision this will inform: [DECISION]

Before writing any SQL or code, design the analysis:
1. Reframe the question into specific, testable sub-questions
2. Define the metrics (with exact formulas, including edge cases)
3. Specify the analysis units (per user? per session? per cohort?)
4. List the data joins and transformations needed
5. Identify possible confounders or data quality risks
6. Recommend the visualization for each sub-question
7. Define what "the answer is X" would look like vs. "the answer is Y"
8. Identify the 3 most likely places this analysis could mislead the decision

End with a 1-paragraph "executive summary template" the team can fill in once results come back.