Back to all articlesContact identity

Decide When Two Contact Records Are Likely to Be the Same Person

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

Two contact records share a name and phone number. One conversation concerns a household purchase; another concerns procurement for a company. Combining them could mix two people’s histories. Rejecting the match because the roles differ could also be wrong: one person may contact your business in different capacities.

To identify duplicate CRM contacts, assess combinations of evidence rather than treating matching fields as proof. Propose a match when evidence connects the records to one person and material contradictions have been resolved. Require clarification when the connection is incomplete or disputed. Keep records separate when reliable evidence establishes different people.

This method covers candidate assessment only. It does not cover executing merges, validating phone numbers, or implementing automated matching. All cases, dates, record references, and quantities in the examples are hypothetical.

1. Give reviewers three decisions

A binary duplicate decision encourages staff to turn uncertainty into certainty. Give reviewers three outcomes with distinct meanings:

  • Propose a match: Evidence supports the same-person explanation, with no unresolved material contradiction. Document the connecting evidence.
  • Require clarification: A connection is plausible, but missing context or disputed evidence prevents a reliable decision. Keep the records separate while investigating.
  • Keep separate: Reliable evidence supports different people. Record the distinguishing fact so the same pair is not repeatedly reconsidered without new evidence.

Insufficient evidence is not the same as evidence of different people. Two records sharing only a common name belong in clarification, not necessarily in a definitive keep-separate category.

These outcomes are assessments, not permission to combine histories, access rights, or communication preferences. A proposed match still needs whatever separate review your organisation requires before downstream action. Clarification also needs an owner and a specific question; otherwise the queue becomes a collection of guesses waiting for someone to approve them.

2. Separate identity evidence from channel evidence

A name describes a person but may not be unique. A phone number provides a contact route, but that route may serve a household, a team, or different people over time. An employer describes a relationship that can change. Ask what each field actually establishes before counting it as support.

Candidate evidence gives you a reason to inspect a pair. Matching names, similar addresses, and shared numbers can all serve this purpose. They do not necessarily connect the records to one person.

Connecting evidence explains why the records belong together despite differences. A dated conversation supplying a replacement email address, supported by consistent later context, is more informative than two similar addresses.

Distinguishing evidence explains why apparently matching records belong to different people. A conversation identifying two colleagues using a reception line changes the meaning of the shared number.

Check where evidence came from. A CRM field, an imported spreadsheet, and a staff note copied from that spreadsheet are one evidence chain, not three independent confirmations. Also distinguish a person’s statement from a staff interpretation: both may be useful, but their attribution and reliability need separate consideration.

3. Compare candidates in a filled decision table

Use the table to compare the same questions across cases. It is a reasoning aid, not a scoring system: several superficial matches do not cancel a material contradiction.

Hypothetical duplicate-evidence cases: all people, organisations, dates, and conversations are fictional.
Case and observationsInterpretationDecisionNext check
A: Matching names. Two Maya Rao records have different phone numbers and email addresses, with no relevant history.The name creates a candidate but supplies no connection between the records.Require clarification.Inspect original intake context for a supplied contact change or another explicit link. Leave separate if none is available.
B: Different employers. An older Maya Rao record lists North Quay Design; a newer one lists Cedar Workshop. A dated conversation in the older record describes the move and supplies the newer email address. Later context is consistent.The documented transition explains the employer difference and connects the records.Propose a match.Reference the linking conversation and inspect later history for contradictory evidence.
C: Shared household number. Two Alex Morgan records share a name and number. A conversation explicitly identifies a parent and adult child using both.The matching fields are compatible with two distinct people.Keep separate.Record the supplied relationship as the distinguishing fact; do not invent labels or suffixes.
D: Shared business number. Priya Shah and Omar Khan share an employer and number. Their conversations identify them as different colleagues using reception.The number identifies a shared business channel, not one person.Keep separate.Record the shared-channel context so future reviewers do not treat the number as personal identity evidence.
E: Direct connection plus contradiction. Daniel Brooks records R-041 and R-118 share a number. A 6 March conversation on R-041 supplies the email on R-118 as Daniel’s replacement address. A 12 March note on R-118 explicitly identifies two different people named Daniel Brooks sharing an office line. The original conversation behind that note has not been located.The email statement supports a match, but the separate-person statement materially contradicts it. The latter’s attribution remains unresolved.Require clarification; block the proposed match.Locate the original 12 March conversation and establish who supplied the replacement address. Propose only if the contradiction is reliably resolved; keep separate if the two-person account is confirmed.
F: Changed number holder. An older record belongs to Elena Cruz. A later conversation identifies the current number holder as Noah Bell and explicitly distinguishes Noah from Elena.The shared number does not establish continuity of identity.Keep separate.Preserve the dated distinction. Route contact-detail corrections through a separate process.
G: Name variation. Samira Patel and Sam Patel use different emails. An older conversation supplies the newer address and identifies Sam as the preferred name. Later context is consistent.A direct explanation connects both the name variation and address change.Propose a match.Cite the connection and check for unresolved distinguishing evidence.
H: Copied corroboration. Two records share a name, employer, and number. A spreadsheet and staff note repeat them, but both derive from one intake submission. The second record has no independent history.Repeated information is not independent corroboration. Duplicate entry is plausible but not established.Require clarification.Inspect record origins. Evidence that the same submission was entered twice, without contradictory history, could support a proposed match.

4. Work through a material contradiction

Case E is different from two conversations showing personal and business roles. Different roles can coexist in one person. An explicit statement that there are two separate people cannot be explained away merely by observing matching fields.

In this hypothetical review, the 6 March conversation directly links R-041 to the email on R-118. That would otherwise support a proposed match. However, the 12 March note says the records concern separate people. Because its underlying conversation is missing, the reviewer cannot yet determine whether it is accurate, misattributed, or an unsupported interpretation.

The immediate decision is require clarification. Neither selecting the newest entry nor counting the matching fields resolves the conflict. The reviewer should locate the source behind the note and examine the context of the earlier address change.

  • Evidence supporting a later proposed match: The original conversation shows that the two-person note was attached to the wrong candidate pair, and the properly attributed address-change conversation still connects these records. Document why the contradiction no longer applies.
  • Evidence supporting keep separate: The original conversation reliably identifies two colleagues with the same name and confirms that the supplied email belonged to the other colleague rather than replacing the speaker’s address.
  • Evidence supporting continued clarification: The original conversation remains unavailable, or the available messages do not establish who supplied which details.

The rule is a checkpoint, not a negative score: an unresolved material contradiction blocks a proposed match. It does not automatically prove different identities. Dates help establish sequence, but recency alone does not establish accuracy.

5. Ask the narrow question without exposing another history

Start with relevant information already available to authorised reviewers: intake context, original conversations, and dated notes. Inspect only what is needed to resolve the identity question. Avoid copying entire histories into an assessment note.

If direct clarification is appropriate, ask for information the person can supply themselves. A hypothetical question might ask whether they previously contacted the business using another email address and, if so, which one. Do not reveal another record’s purchases, employer, or private conversation as a recognition test.

A duplicate-review queue does not itself justify outreach. For WhatsApp contact, recipient permission and opt-outs still matter. Approved templates are used to initiate Platform conversations and outside the 24-hour customer-service window, which opens or resets with a user message. Check the current WhatsApp Business Messaging Policy for precise requirements. Template approval does not guarantee delivery.

If clarification cannot be obtained appropriately, leave the assessment unresolved. An ageing queue is an operational issue, not new identity evidence.

6. Reuse this filled assessment note

The following hypothetical note turns Case E into a handoff another reviewer can challenge. Reuse the field labels, replacing the contents with evidence from the actual pair. This is a recommended operating format, not a claim that a particular product provides these fields or controls.

Candidate pair
R-041 and R-118, both labelled Daniel Brooks.
Supporting evidence
6 March conversation on R-041 supplies the email currently on R-118 as a replacement address. Shared number also observed, but not treated as proof of identity.
Contradictory evidence
12 March note on R-118 identifies two separate people named Daniel Brooks using one office line. Underlying conversation not yet located; attribution unresolved.
Shared-source warning
The email field on R-118 may derive from the 6 March conversation. Do not count the field and conversation as independent confirmations.
Assessment
Require clarification. Records remain separate; no proposed match while the material contradiction is unresolved.
Question to resolve
Does the original 12 March conversation concern this pair, and did the 6 March speaker supply their own replacement address or a colleague’s address?
Owner and next action
Assigned reviewer Morgan to locate both source conversations before the next scheduled review. If unavailable, retain the unresolved assessment and record the missing evidence.
Reopening conditions
Propose a match only after reliable attribution resolves the contradiction and preserves the direct connection. Keep separate if reliable source context confirms two people.

A strong note separates observation from interpretation. It also tells the next reviewer what would change the decision, rather than merely stating that the records look similar.

7. Reopen decisions when evidence changes

If someone challenges a proposed match before downstream action, pause the proposal and inspect the disputed evidence. If someone challenges a keep-separate decision, ask what new connection changes the earlier reasoning. Repeating the same field comparison is not a fresh assessment.

Keep the earlier rationale alongside the revision so reviewers can see what changed. A corrected attribution should identify which evidence no longer applies and why; it should not silently erase the previous concern.

When records have already been combined or another operational action has occurred, flag the issue to the responsible owner. Recovery belongs to a separate correction process. Candidate assessment should not become an improvised attempt to reconstruct histories or restore preferences.

8. Check whether the rule produces reviewable decisions

As a hypothetical exercise, select 20 suspected pairs covering shared numbers, employer changes, copied evidence, and contradictory histories. Have two authorised reviewers assess each independently using the same three outcomes.

Compare their reasons as well as their decisions. Review disagreements about whether evidence is independent, whether a statement is correctly attributed, and whether a conflict is material. Agreement alone does not establish accuracy if both reviewers rely on the same unsupported assumption.

Check observable outputs: every proposed match should have a documented connection and no unresolved material contradiction; every clarification case should name a specific question and owner; every keep-separate decision should state the distinguishing evidence. Use missing explanations to improve the review method, rather than treating the number of proposed matches as a measure of success.

CRMContact identityData qualityDuplicate assessment

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications