You are asked how you would handle a Bigbear dashboard where the same KPI shows different numbers across different source tables or models. Explain how you would use SQL to reconcile the sources, identify whether the mismatch comes from joins, aggregations, filters, missing rows, duplicate rows, or inconsistent metric definitions, and describe how you would confirm the final fix.

A strong answer should be operational, not theoretical. The interviewer expects a step-by-step debugging workflow and the SQL patterns you would use, not just a high-level statement that you would “check the data.”