How do you clean up duplicate data?

Duplicates are identified using reliable keys when available - tax ID, SKU, ID - and with matching rules when the data is messy. We define which record survives, how fields get merged, and what needs manual review. Deduplication is tested before any information is deleted or overwritten.