Sometimes the business reports differ from the source system. How would you investigate the mismatch? What steps or areas would you examine?

First, isolate the scope by comparing row counts between source and target for the affected period. Next, sample data to identify specific mismatched rows. Use Snowflake queries to join the source and target on the natural key and flag differences in key columns. Check ETL logs for errors or transformation failures. Validate transformation logic by re‑running the transformation on a subset of source data locally and comparing results. Verify that the source data itself is clean—look for nulls, data type mismatches, or duplicate keys. Examine any incremental load logic to ensure that only changed rows are processed. If the mismatch persists, review the Snowflake role permissions and data masking policies that might alter visible data. Finally, document findings, fix the root cause (e.g., adjust transformation, correct source data, or modify load schedule), and re‑run the pipeline to confirm resolution.