Scenario and Case Style

21. You've been asked to build a retention cohort analysis with no existing template. Walk me through your approach, from question to final chart.

A strong answer covers defining "active," choosing the cohort unit (weekly vs. monthly), structuring the query, and how they'd sanity-check the resulting curve.

22. A stakeholder asks you to "just pull the numbers" on a vague question. How do you handle the request?

Look for the instinct to translate vague to precise before writing any SQL, referencing something like the Lesson 4.2 and 4.3 frameworks, without needing to name them explicitly.

23. You run a query and the result is exactly zero, when you expected a meaningful number. Walk me through how you'd debug it.

Tests the specific, common failure modes: date filtering against a timestamp, a join that excluded everything, a typo in a filter value, checked in a sensible order.

24. How would you design an experiment analysis query to compare a test group and a control group fairly?

Look for consideration of comparable time windows, consistent metric definitions across both groups, and awareness of what SQL alone can and can't tell you about statistical significance.

25. Design, at a high level, the SQL and metrics you'd use to monitor the health of a feature you've shipped in the past. Walk me through your key decisions.

This is the free-form capstone-style question. A strong answer names the specific metric and its precise definition, the comparison baseline, how they'd catch a misleading number, and what decision the ongoing monitoring is meant to support, demonstrating the whole course's thinking in one scenario.