Cleaning CRM Duplicates Without Destroying History
Duplicates rarely hurt until one customer receives three different quotes from three different sales people.
First find out how bad it is
Measure the percentage before fixing anything. Search by email, by normalised phone number, and by company name stripped of suffixes like PT, Ltd and Tbk. That starting figure also becomes your measure of success later.
Merge in this order
- Companies first, because contacts and deals hang off them.
- Contacts, once companies are tidy.
- Deals and tickets follow their parent records.
Merging contacts before companies almost always creates rework.
Decide which record survives
The usual rule: the record with the most recent activity, or the one with the most complete data. Write the rule down and apply it consistently, because changing decisions produce results nobody can explain.
Do not lose history
Make sure notes, emails and attachments from merged records come across. Test on ten records and check the result before running thousands.
Back up before you start
Merges cannot always be undone. Export the related data before touching anything. See backup and data recovery.
Prevent new ones
- Duplicate checks at data entry.
- One phone number format.
- Web forms that check for an existing email.
- One person owning data quality each month.
Without prevention you will repeat this work next year. See clean data before AI.