Resolve Duplicate Customer Records Safely

ROLE Act as an entity-resolution analyst. CONTEXT Review [CUSTOMER DATA FILE] using [AVAILABLE IDENTIFIER FIELDS]. The organization’s merge rules are [MERGE POLICIES], and the relative costs of false merges versus missed matches are [ERROR PRIORITIES]. INSTRUCTIONS Define exact-match and fuzzy-match stages. Normalize comparison fields while preserving their originals. Explain which combinations provide strong evidence of identity and which are weak, such as shared addresses or common names. Propose blocking rules to limit comparisons and a transparent match-scoring approach. Separate high-confidence matches, manual-review candidates, and records that should remain distinct. Address transitive matching risks and specify how conflicting field values should be resolved. Recommend a small labeled review sample to evaluate proposed thresholds before applying them widely. CONSTRAINTS Avoid fabricated statistics, match accuracy, or duplicate counts. Flag assumptions. Do not expose unnecessary personal information or automatically merge ambiguous records. OUTPUT FORMAT Provide a matching specification, candidate-pair review template, survivorship rules, and reversible implementation steps. Include a merge audit log and validation checks for record counts, retained identifiers, and downstream relationship integrity.