Stakeholder Data Discrepancy

A stakeholder tells you the numbers look wrong, what do you do?

Asked in the loop rounds 3 to 4 as a scenario-based question. You are given access to Amazon QuickSight dashboards, Amazon Redshift tables, and the underlying event logs. The stakeholder says a key KPI changed sharply overnight, but they cannot tell whether it is a data issue, a metric definition issue, or a real product shift.

Requirements

1. Describe the first 3 checks you would run to validate the metric.
2. Explain how you would decompose the KPI to find the source of the anomaly.
3. List the specific data sources you would compare and what each would confirm.
4. State how you would communicate findings and next steps to the stakeholder.

Data available

• dashboard_snapshot table, daily KPI outputs by segment
• event_logs table, raw product events with timestamps and user_id
• metric_definition table, business logic and filters for each KPI
• pipeline_run_log table, ETL status, row counts, and freshness by job