How to find duplicate contacts in a CRM
Learn to find duplicate contacts in a CRM, verify which records describe the same person, and propose merges that preserve field conflicts and opt-outs.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What it means to find duplicate contacts in a CRM
To find duplicate contacts in a CRM, look for records that refer to the same person and gather evidence strong enough to justify grouping them. A contact may appear twice because different teams entered their details or because an import created a new record. CRM deduplication starts with contact matching and ends with a reviewable proposal that names the records, explains the identity decision, and shows what data would survive.
A customer relationship management system, or CRM, can hold different kinds of history on each copy of a contact. One record might have the current job title and another the unsubscribe. A responsible merge plan records the surviving ID and uses a documented rule for each field. It also preserves the untouched records and gives an owner a way to resolve ambiguous cases before applying changes.
What it does for the business
- Give account teams a coherent contact viewA reviewed plan lets the CRM owner consolidate confirmed duplicates while keeping the evidence behind important fields. Reps can spend less time looking through copies to find the relevant contact record.
- Protect contact preferencesShowing exactly which stops survive gives operations a way to verify the resulting record before it enters an active list. This also makes disagreements between subscription sources easier to investigate.
- Keep corrections manageableA mapping from each duplicate ID to a proposed survivor lets an admin review individual groups and find the old values. If a decision is challenged, the team can revisit the group with the original evidence in hand.
- Find where duplicate creation beginsPatterns in the proposed groups can point to an import or entry path worth examining. Operations can investigate those paths and adjust how new contact records are matched before the next import.
How to find duplicate contacts in a CRM, step by step
Define scope and save the current records
A merge decision needs stable record IDs and a source you can recover.
What to do
- Decide which contact objects, business units, or source exports the review covers. Record the snapshot date and reconcile the number of records with the export scope. Use the current approved export when multiple versions are available.
- Keep an unchanged copy of the source. Capture stable contact IDs, email, name, company, and relevant phone and subscription fields. Keep the source of each field and any timestamps that explain when it was observed or changed.
- Check whether the destination has separate records for people, shared addresses, or contact roles. Learn which IDs carry open work or important relationships so you can evaluate the effect of choosing a different survivor.
- Document source gaps before profiling the data. If part of the contact history is unavailable, record which decisions can proceed and which need more evidence.
Why
A partial export can make one copy look like the only record for a person. An old export can make a superseded value look current. A dated scope and preserved source give you a basis for deciding which merges can be reviewed now.
Find candidates and test whether the records match
A candidate match is a question you can answer with record evidence.
What to do
- Normalize spacing and domain case in a comparison copy and retain raw values. Document how you treat local address case and punctuation. Leave mailbox aliases intact unless you have evidence that the address rules support the equivalence.
- Group exact normalized emails as candidates. Check names, companies, phone fields, and any reliable person identifiers to detect shared addresses, recycled addresses, or company conflicts.
- Inspect a shared mailbox separately. An address used by a purchasing team may appear on several contact records for valid reasons. Look for evidence that ties each record to the same person before grouping it.
- Record which pair of records matches, which values support the match, and which values conflict. Leave a similar name in the review queue if email or other reliable identity evidence is missing.
Why
A false match can connect activity and preferences to the wrong person. A pairwise evidence record lets a reviewer decide which link is sound. It also exposes weak links that would otherwise pull unrelated contacts into a larger group.
Choose the survivor and resolve fields one at a time
The best ID to keep and the best source for a value can come from different records.
What to do
- For each supported group, select the survivor record, the ID the proposed merge keeps. Consider continuity of active work, trusted identifiers, and important relationships. List every other member with the rule and evidence that place it in the group.
- Set field precedence, the rule that decides which source contributes each value. Use a reliable field timestamp where recency matters. If the source only shows a general record update, check whether it actually dates the field you are comparing.
- Show both values when company, title, or other material fields conflict and the evidence leaves the choice open. Keep the relevant stop flags and unsubscribe history visible. Flag disputed consent timing for review.
- Check the group for chains of weak matches. If two records share a company and a third has a similar name, require a supported identity link before including the third. Keep the remainder outside the proposed changes.
Why
A rule that always keeps the newest row can discard a reliable ID or a suppression entry. Choosing the ID and field values separately makes each choice reviewable and gives an operator an explicit way to resolve conflicts before a change.
Check the proposal against every source ID
Every contact ID must appear in exactly one merge group or in the untouched remainder.
What to do
- Compare the input IDs with the combined member and remainder IDs. Look for missing records, repeated membership, or IDs outside the source. Check that every survivor belongs to the group it is meant to represent.
- Review shared addresses, missing timestamps, and groups with conflicting company data. Read the evidence for groups formed from several links and confirm the links support one identity throughout.
- Trace every unsubscribe and suppression to the proposed result. Compare the row evidence with the retained restriction. Leave a hold where you cannot establish the sequence or scope of conflicting subscription events.
- Have a reviewer reproduce the match and field decisions from the saved source. Record open issues with the group IDs so an admin can exclude those groups from a later change.
Why
A table can appear internally consistent while dropping an input ID or relying on a faulty link. Checking the full membership and the most consequential fields makes the proposal safer to use in a later administrative review.
Hand the reviewed proposal to the CRM owner
A merge proposal is ready for review when the operator can explain and trace every change.
What to do
- Deliver the proposed groups, untouched records, and conflict list with the source date and scope. Include the survivor reason and retained restrictions so reviewers can see the consequence of each decision.
- Ask the CRM owner to verify how their system handles linked activity and subscriptions. Before they merge duplicate leads or contacts, have them compare current values with the snapshot and decide how to preserve relevant associations and recover old values.
- Give the owner a way to record decisions on held groups. If the source changes while a group is under review, recheck identity and field choices using the refreshed records.
Why
The candidate review describes what the evidence supports at a particular date. The CRM owner applies that proposal in a system with its own merge rules. A dated handoff gives the owner enough information to spot drift before committing a change.
Common mistakes
- Grouping by name and company aloneRequire identity evidence before combining contacts who share an employer and similar names.
- Assuming an address belongs to one personCheck for shared and reassigned mailboxes before treating an email match as confirmed.
- Keeping every field from one rowSelect each field by its own evidence and keep unresolved conflicts visible.
- Dropping a stop during a mergeTrace every applicable suppression into the proposed result.
- Trusting a balanced count aloneCheck unique identifiers so a repeated membership cannot hide a missing row.
What a good result looks like
- The source scope and snapshot are clear.
- Every proposed group has a survivor, all member IDs, and identity evidence.
- Field choices have reasons and unresolved conflicts are visible.
- Suppression follows the contact and unknown timing remains unresolved.
- Each input ID appears exactly once among the proposed groups and untouched records.
And then what? Using the result
- Have the CRM owner review the proposed groups alongside the current records before applying changes.
- Assign unresolved groups to the people who can confirm identity or field history.
- Check the CRM after any approved change to verify activity links and suppression on the surviving records.
- Use recurring duplicate patterns to inspect the source imports or data entry rules that created them.
When to run it
- After a CRM migration or combined importReview duplicates using the retained source IDs. Decide what should survive before consolidating data that came from separate systems.
- When account owners see competing contact recordsInvestigate each copy's identity and activity before choosing where the team should record new work.
- Before transferring contact preferencesUse the merge plan to preserve stop flags and identify conflicts that need subscription evidence.
- When duplicate records keep being createdCompare candidate groups by source so operations can inspect the process where the records enter.
Questions people ask
How do you find duplicate contacts in a CRM?
Export the current records with their IDs, group exact normalized email matches as candidates, and check each against corroborating and conflicting evidence. Record the supported groups and account for every untouched record.
Is the same email address enough to merge contacts?
A matching email is a candidate signal that needs context. A shared mailbox or a company conflict calls for evidence that both records represent the same person.
How do you choose a survivor record?
Keep the ID that best preserves reliable identity and continuity of active work. Select individual values using documented field precedence and flag unresolved conflicts.
What happens to opt-outs when contacts are merged?
Retain applicable opt-outs and suppression on the resulting record, with their source and scope. Flag disputes about chronology for review before treating the record as eligible for outreach.
How do you validate a contact merge plan?
Reconcile every input ID, reproduce group membership from the evidence, and check field winners and suppression against the raw records. Review shared mailboxes and ambiguous timestamps explicitly.
Can I use the merge plan to update the CRM?
You can give the local proposal to your CRM owner for review. The owner checks the current records and the CRM's merge rules before applying any change.
More jobs in Leads
Find duplicate contacts in a CRM, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac