Cleaning and importing customer records
Prepare a data dictionary
Before importing, define each column, its type and whether it is required. Distinguish companies from people and phone numbers from account identifiers. Preserve leading-zero values as text and inspect country codes instead of adding the same code to every record.
Identify the authoritative source for email, owner and status. When records conflict, the newest file is not necessarily correct: its export date may be recent while its contents are old. Preserve source IDs and modification dates when available.
Verify identity before merging
Treat email and phone as matching signals rather than universal proof of identity. Two employees may share an address, and a number can change owners. Review ambiguous records containing active deals or important history before merging.
Start with a representative sample: long Arabic names, empty fields, varying date formats and multiple numbers per person. Inspect the imported contact and linked deal, checking ownership and notes rather than only the imported row count.
Make quality an ongoing process
Retain the source file, mapping decisions and rejected-record log. Do not repair an import by uploading everything again without understanding duplicate prevention. Fix the failed subset and reconcile totals and relationships afterward.
Monitor decision-relevant fields such as contact method, owner and next action. Investigate why users leave values empty: the field may be unclear or unavailable at entry time. Removing a field that supports no decision may improve quality more than another warning.