How to clean up duplicate contacts in a CRM

Why this matters

Duplicate contacts split the context a sales or support team needs. One record holds the work email and open deal, another holds the direct phone and last call note, and a third holds the billing address. Reps pick the wrong record, follow up twice, or hand a deal to finance with missing details. How to clean up duplicate contacts in a CRM is mostly a data hygiene job, but it also protects the trust people place in the system.

Duplicates also distort reports. Contact counts, source attribution, owner lists and pipeline totals become unreliable when the same person appears in several records. A forecast built on split records hides the real stage and the real next step. Support sees a different history from sales. The cleanup is not cosmetic.

The risky part is the merge. A merge combines fields and related records, and the wrong survivor can overwrite a valid phone number, drop a note, or move an open deal to the wrong owner. Treat the work as a controlled data project, not a bulk button press.

Before you start

Start with a full export of contacts and every object that links to them: deals, tickets, notes, activities, email subscriptions and invoices. Capture record IDs, created dates, last activity and owners. Pause imports, web forms and sync jobs for the cleanup window, or record the cutoff time so you can reconcile changes afterward. Test the export before you merge anything.

Define your match keys before you sort the list. Exact email after lowercasing is a strong key. Normalized phone with spaces and punctuation removed is another. Company domain plus first and last name catches many business contacts, but it also catches shared inboxes and common names. Add address and company record ID as secondary checks. Decide which matches are safe to merge and which need review.

Write a survivor policy. The surviving record might be the one with the most complete history, the correct owner, or the most recent activity. Pick one rule and apply it consistently. Decide who can approve merges and who can only suggest them. Check the integrations that create contacts: website forms, email tools, support desks, billing systems and calendar connectors. If one writes a new contact for every reply, fix that before you clean the list.

How to do it

Run a duplicate report or export the contact list and group records by your match keys. Score each group as high, medium or low confidence. High confidence means exact email or phone plus the same person. Low confidence means shared company domain and a common name. Start with high confidence groups, because a wrong merge there is less likely and the pattern teaches you what to expect.

Choose the survivor for each group and write it in a merge sheet. List the fields to keep, the fields to fill from the loser, and the related records that must move. Merge in small batches, then open the survivor and check the deal links, notes, owner, subscription status and last activity. If the CRM offers field-level merge choices, use them. Do not accept the default survivor blindly.

Fix the intake paths after the first pass. Add duplicate rules on create and edit for exact email and phone, with a warning for probable matches and a block for exact ones. Clean your import templates and map external IDs so a sync updates an existing contact instead of creating a new one. Give only administrators the merge permission, and log every merge with a reason. The steps below put this into order.

1. Freeze the intake paths. Pause forms, imports and sync jobs that can add contacts during the cleanup. Note the time so you can replay or reconcile changes after the merge window closes.
2. Export contacts and related records. Export contacts with record IDs, owner, created date, last activity and all match fields. Include related deals, tickets, notes and subscriptions so you can verify what moves during a merge.
3. Build duplicate groups with match keys. Normalize email to lowercase, strip spaces from phone numbers and compare company domains. Group records that share a strong key, then review weaker name and address matches by hand.
4. Pick the surviving record for each group. Choose the record with the most complete history and the correct owner, not simply the oldest. Write the choice in a merge sheet before you touch the CRM.
5. Merge in small batches with field rules. Merge ten to twenty groups at a time and check the result after each batch. Copy missing values into the survivor and confirm that open deals, notes and subscriptions move with it.
6. Update source systems and integrations. Fix the form, import template, email tool or support connector that created the duplicates. If the source still sends dirty data, the cleanup will not last.
7. Add create and edit duplicate rules. Turn on duplicate rules for manual entry and imports, with a warning or block for exact email and phone matches. Add a merge permission for admins only.
8. Document the rules and train the team. Write down the match keys, survivor rules and who can merge. Show reps how to search before creating and how to request a merge.

How to check it worked

Re-run the same duplicate report after the sync delay. Compare group counts by confidence and open a sample of merged records. Confirm the fields, owner, open deals, notes and subscription status landed on the survivor.

Test the intake paths. Create a contact with an email that already exists and see whether the warning appears. Import a small file with a known duplicate and check that the CRM matches the existing record. Let one integration run and confirm it updates rather than creates.

Review the reports that depend on contacts: owner lists, source counts, email audiences and the handoff to finance. A merge that looks correct on the record can still leave a broken report if a related object did not move. Schedule a second pass after the next large import.

Mistakes to avoid

Do not auto-merge a whole list without review. Matching only on email misses people who use a work and personal address, and matching only on name merges different people with common names. Deleting a duplicate is not a shortcut if the record holds notes, deals or unsubscribe flags. Merge to keep the history, and delete only empty test or spam records.

Do not ignore related records and integrations. A contact merge can move an open deal to the wrong owner or break a support ticket link. Check the objects that point at contacts before you start. Keep an audit trail of survivor choices and merge reasons so you can explain a change later.

Do not treat the cleanup as a one-time event. New forms, imports and connectors will recreate the mess unless you add rules and review the duplicate report on a schedule. If the cleanup is part of a wider CRM reset, one option among others is the Sales CRM implementation package from Cloudverg at a $9,500 fixed price, which includes a contact, company and deal model without duplicate mess.

Frequently asked questions