guided-investigation

To structure an open-ended data investigation, decompose vague questions into 3-5 specific sub-questions with explicit success criteria, then map them to available data before executing targeted SQL queries.

The best way to track data discovery progress is using a formal five-phase workflow that records rationale, methods, and results for each sub-question, ensuring systematic investigation and preventing lost insights.

To synthesize SQL query results into actionable recommendations, integrate findings from individual sub-questions into a coherent narrative, capturing observations and data quality checks to support final conclusions.

Yes, you can document data gaps and quality checks during the data mapping phase by aligning your decomposed sub-questions with available data sources and explicitly recording missing information.

Yes, the framework supports optional scripts, references, and assets for extensibility, allowing you to enforce YAML frontmatter while customizing the investigation workflow to fit specific data analysis needs.

Use structured question decomposition when tackling broad, open-ended analytical problems like identifying why user engagement is dropping, ensuring your data discovery process remains trackable and produces actionable conclusions.