Data Cleansing Defect Log Worksheet
Data cleansing needs a backlog, owners and a way to size the effort. This defect log records each issue by domain, severity and owner, and totals the estimated hours and cost to fix it.
| Defect ID | Domain | Issue | Records affected | Severity | Owner | Found on | Minutes per record (min) | Hourly rate | Fix hours | Fix cost | Status | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
11 | $440 | |||||||||||
11.5 | $460 | |||||||||||
14.17 | $779 | |||||||||||
10.33 | $362 | |||||||||||
13.2 | $726 |
Log a defect once per root cause, not per record, and keep the list of affected record IDs in an attached file. Fix hours and cost use records times minutes per record, and are estimates for planning.
Assisted data cleansing
Want help proposing fixes at scale with approval in the loop? Request an expert review.
Optional. The tool and every export work without this. No spam.
How to use the log
Create one row per root cause. Describe the issue plainly, such as missing unit of measure on purchased items, and enter the number of affected records from your rule run. Set severity by business impact, name an owner, and estimate minutes per record. The worksheet calculates fix hours and cost, so you can plan cleanup sprints before a migration or audit. Update status as work moves from open to closed.
- One row per root cause
- Severity reflects business impact
- Estimate minutes by sampling a few records
- Verify fixes with a rerun of the rule
Setting severity
Critical defects cause financial loss, compliance breach or production stoppage, for example duplicate vendors with the same bank account or missing export classification on controlled parts. High defects break a process regularly, such as missing planning attributes. Medium defects cause rework. Low defects are cosmetic. Agree the scale with the data owners and apply it consistently.
Using it for migration
Before an ERP migration, a defect log is the usual way to track cleansing in source data. Fix what you can at the source, not in the load file, so the clean data persists. Use the cost total in your project plan, and track closure rates in each mock load. Carrying known defects into the new system is a decision to make explicitly, not by default.
Common mistakes
Logging every record as its own defect, fixing in the load file only, closing without verification and ignoring the root cause. If the same defect returns each month, add a validation at data entry or a rule in your catalog.
Frequently Asked Questions
What is a reasonable minutes per record?
Sample ten records and time the fix. Simple field fills may take one to three minutes. Merging duplicates, which needs checking of open documents, often takes ten to twenty.
How do I prioritize the backlog?
Sort by severity first, then by records affected and effort. Fix critical defects regardless of cost, then pick high impact, low effort items for quick wins.
Should I log defects found in testing?
Yes if they are data defects. Tag them with the test cycle in the issue text so you can see whether the count falls cycle by cycle.
Does cost include system time?
No. It is labor only. Add costs for tools, consulting or downtime in your project budget.
If you have a large backlog, we can build agents that propose fixes for human approval inside your ERP, such as standardized descriptions and duplicate merges.
Related Tools
Master Data Quality Rules Catalog Worksheet
An editable catalog of data quality rules with check type, threshold, owner and run frequency.
ERP OperationsERP Data Quality Scorecard Worksheet
Enter record and pass counts per master data domain and the worksheet calculates quality percentages and an overall score.
Supply Chain & InventoryInactive Item and Obsolescence Review Worksheet
Review items with no recent movement, decide to keep, deactivate or write down, and calculate extended value and reserve.
Costing & FinancePoor Master Data Quality Cost Calculator
Estimate the annual cost of master data errors from rework time, expediting and stewardship, and see how it scales with error rate.