CDS Troubleshooting & Problem Resolution 2 — Questions and Answers
Question 1: A data steward discovers that two systems report different customer counts for the same reporting period. What is the FIRST step in resolving this discrepancy?
- Declare one system authoritative and discard the other
- Identify the business rules and transformation logic each system uses to count customers (Correct answer)
- Average the two counts and use that as the official figure
- Escalate immediately to the CIO without further investigation
Correct answer: Identify the business rules and transformation logic each system uses to count customers
Understanding each system's counting logic and business rules is essential before determining the source or resolution of a discrepancy.
Question 2: Which technique is most effective for tracing a data quality error back to its point of origin in a complex data pipeline?
- Data profiling at the target only
- Data lineage analysis (Correct answer)
- Re-running all ETL jobs from scratch
- Asking end users which record looks correct
Correct answer: Data lineage analysis
Data lineage analysis tracks data from its source through every transformation, making it the best tool for pinpointing where an error was introduced.
Question 3: A business user reports that a critical KPI dashboard shows NULL values where numbers are expected. After investigating, the data steward finds the issue is in the ETL join logic. Who should be engaged to resolve this?
- Only the business user who reported the issue
- The data steward alone by rewriting the dashboard query
- The ETL developer and the data steward collaboratively (Correct answer)
- The database administrator only
Correct answer: The ETL developer and the data steward collaboratively
ETL join logic fixes require collaboration between the ETL developer (technical fix) and the data steward (validating business requirements).
Question 4: A dataset used for regulatory reporting is found to contain duplicate records. Which approach best prevents recurrence after removing the duplicates?
- Manually review all records each quarter
- Implement a deduplication rule and unique constraint at the point of ingestion (Correct answer)
- Change the reporting tool to ignore duplicates
- Archive the duplicate records without investigation
Correct answer: Implement a deduplication rule and unique constraint at the point of ingestion
Implementing preventive controls at the ingestion point stops duplicates from entering the system rather than treating them after the fact.
Question 5: When performing root cause analysis on a data quality issue, a fishbone (Ishikawa) diagram is most useful for:
- Documenting data lineage across systems
- Identifying and categorizing potential causes of a problem (Correct answer)
- Tracking the number of defects over time
- Defining SLAs for data delivery
Correct answer: Identifying and categorizing potential causes of a problem
A fishbone diagram organizes potential causes of a problem into categories, facilitating structured root cause analysis.
Question 6: A data steward finds that date fields are formatted inconsistently across source systems (MM/DD/YYYY vs. YYYY-MM-DD). Which resolution strategy addresses the root cause?
- Create a view that converts dates at query time only
- Enforce a standard date format in the data dictionary and update ingestion pipelines to conform (Correct answer)
- Ask users to manually convert dates before uploading
- Document the inconsistency and take no further action
Correct answer: Enforce a standard date format in the data dictionary and update ingestion pipelines to conform
Standardizing the format in the data dictionary and enforcing it at ingestion eliminates the root cause rather than applying band-aid fixes downstream.
Question 7: After resolving a recurring data issue, what should a data steward document to prevent future recurrence and support organizational learning?
- Only the final corrected records
- A post-mortem including root cause, resolution steps, and preventive controls implemented (Correct answer)
- An email to the team summarizing what was wrong
- Nothing — verbal communication is sufficient for minor issues
Correct answer: A post-mortem including root cause, resolution steps, and preventive controls implemented
A post-mortem provides institutional knowledge that helps teams recognize and prevent similar issues in the future.
CDS Troubleshooting & Problem Resolution 2
A data steward discovers that two systems report different customer counts for the same reporting period.
What is the FIRST step in resolving this discrepancy?