Data Quality, Cleansing and Mapping Rules

Keep this: job aid

Migration data quality card

• Profile every extract with the same scripts: completeness, validity, uniqueness, consistency, accuracy, timeliness.
• Profile against the target template and configuration, not only the source.
• Triage findings: load blocker, business quality issue or selection question.
• Every cleansing rule names the records, the action, where it applies and the approving owner.
• Cleanse at source first; transformation for mechanical rules; target fixes only as fallback.
• Merges need survivorship rules, re-pointing and an old-to-new cross-reference.
• Mapping specification: one row per template field, a rule type for each, unmapped values reported not defaulted.
• Report per object: open findings, rule pass rate, error rate with in-scope count, reconciliation and owner findings.
• Version every baseline; change through requests with impact analysis and retest.
• Freeze rules before the dress rehearsal; hand them over as ongoing checks after go-live.