In this guide 6 sections
Duplicates come from web forms, spreadsheet imports, two reps adding the same company and phone numbers typed in different formats. Export and back up first. Group matches by email, domain and normalised phone, then merge each group into one surviving record. Finally, switch on matching rules and required fields at every entry point.
Where duplicates actually come from
Start by finding the source before you merge anything. Cleaning without fixing the entry point only buys a few quiet weeks.
Web forms are the first suspect. A visitor fills the contact form twice, or a trade-show list arrives as a CSV and the import creates fresh rows for people already in the system. Then come the reps. One enters “Harbor Tools Ltd”, another enters “Harbor Tools”, and the CRM sees two accounts.
Phone numbers are quieter offenders. The same mobile appears as 555 0142 7788, +1 555 014 27788 and (555) 014-2778. A human sees one number. A database sees three. Matching fails quietly.
- Web form submitted twice or from two campaigns
- Imports from event lists and purchased data
- Two reps keying one company under variant names
- Phone numbers with spaces, brackets and country codes
- Contacts who changed employer and were re-added
Measure the mess before touching it
Export contacts, companies and open deals to CSV. Open the file in a spreadsheet and sort by email, then by company name, then by phone with the punctuation stripped out. Count the clusters.
Sample fifty clusters by hand. Mark each as a true duplicate, a false match or a judgement call. False matches are common with shared inboxes such as sales@ and info@, and with a head office that has many real contacts.
This sample tells you how aggressive the merge rules can be. If forty-five of fifty are true duplicates, automatic merging by email is safe. If many are false matches, review by hand.
A safe merge routine
Back up the whole database first and confirm the export opens. Merges in most CRMs cannot be undone, so the backup is your only way back.
Pick a survivor rule in advance. The oldest record usually keeps its ID, because reports and email history point to it. Fields are filled from the other copy only where the survivor is blank. Activity notes, open deals and attachments move across.
Merge in small batches of a few hundred, check the deal totals and owner assignments after each batch, and log what changed. Never merge accounts and contacts in one pass. Do accounts first, because contacts hang beneath them.
Matching rules and required fields that stop it returning
Most CRMs can warn on a duplicate at the moment of entry. Turn that on. Match contacts on email, accounts on website domain, and treat phone as a supporting signal once stored in one format.
Make a short list of required fields. A lead without a source, an owner and a valid email or phone is a lead nobody can follow up. Use picklists for country, industry and lead source, so “New York”, “NYC” and “N.Y.” stop competing.
Set a formatting rule that stores phone numbers in international form. Apply the same rule to web forms and imports, not only to manual entry.
- Email as the contact match key
- Website domain as the account match key
- Phone stored in one international format
- Lead source, owner and region as required fields
- Picklists instead of free text for city and industry
Keep it clean with a monthly habit
Give one person ownership of data quality. Their job is small: run the duplicate report, review the flagged pairs and chase reps whose records keep failing.
Add a saved view for leads with no owner or no source. Review it every month. Add another for contacts whose emails have bounced.
Before the next import, run the file through the same normalising steps and a match against existing records. Reject rows that fail rather than fixing them after the fact.
CRM duplicate cleanup tracker
The sheet opens with a source list covering web form, import, rep entry and phone format, each with a fix and an owner. Next comes a merge log with columns for cluster, survivor ID, merged IDs, batch date and check result. A closing section names the matching keys and required fields.
Open a blank worksheet to printQuestions owners ask
Should I delete duplicates or merge them?
Merge. Deleting loses the notes, deals and email history attached to the copy. A merge keeps one record and moves the activity across, so reports stay accurate.
How do I handle shared emails like info@ or sales@?
Exclude them from automatic matching. Flag them for manual review and match those contacts on name plus company instead, since several real people may use one inbox.
Can my CRM find duplicates on its own?
Most mid-range CRMs include a duplicate finder and entry-time warnings. Check the settings, then confirm the match keys suit your data before enabling automatic merges.
How often should I run a cleanup?
Review flagged pairs monthly and run a full audit once a year or before a big import or migration. Frequent small reviews are cheaper than a rare large rescue.