Back to all articlesContact identity

Decide When a Contact Merge Needs a Second Reviewer

Alactic Editorial9 min read

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

Consider two hypothetical cases. One staff member finds two near-empty records created from the same intake submission. Another finds matching phone numbers on records with different names, separate conversations, and conflicting identity claims. Sending both through the same approval queue wastes attention on the first and may not provide enough scrutiny for the second.

Allow a single authorised operator to merge only when identity evidence is clear, linked history is simple, recovery is understood, and the consequences of error are limited. Require independent review when those conditions are not met. If identity remains disputed, a second signature is not permission to guess.

This contact merge approval process assigns authority according to evidence and consequences, not the operator’s confidence. It is a recommended operating design, not a claim that any particular software provides these controls. All cases and illustrative quantities below are fictional.

1. Separate identity judgment from permission to act

A merge involves two decisions: whether the records represent the same person, and whether combining them is acceptable under a specific handling plan. A strong answer to one does not settle the other.

Two records may clearly belong to one person but contain histories that would be difficult to disentangle. Conversely, two empty records may be easy to combine while representing different people who share a household phone.

Use three routes:

  • Single-operator authority: An authorised operator may proceed within the defined low-risk boundary and record the evidence.
  • Independent second review: Another authorised person checks the identity judgment, affected history, and proposed handling before execution.
  • Hold for resolution: Nobody proceeds until a specific evidence gap or handling problem is resolved. Management approval cannot substitute for missing identity evidence.

Approval applies to the exact record pair, surviving values, linked-history treatment, and recovery account. Changing those details materially requires reconsideration. A reviewer’s approval authorises that plan; it does not imply automatic execution.

2. Assess four dimensions without averaging away a blocker

Identity uncertainty

Record both supporting evidence and contradictions. Matching names, email addresses, or phone numbers are clues, but shared addresses, reassigned numbers, and entry mistakes can mislead.

Stronger evidence explains how both records arose and resolves differences. A documented repeated import from one original submission is different from two records that resemble each other. Several fields copied from one source are not several independent confirmations.

Linked-history complexity

Inspect conversations, open work, organisation relationships, ownership, and preferences. Complexity is not simply volume: a short history containing conflicting instructions may need more judgment than a long, consistent history.

Ask whether consolidation could change attribution, conceal a separate relationship, or make one person’s activity appear to belong to another.

Reversibility

Determine what recovery means in the current workspace: whether records can be separated, which values and relationships could be restored, and how later activity would be handled. A screenshot preserves evidence but does not establish a restoration method.

Distinguish known irreversibility from unknown behaviour. The first may support an explicitly authorised risk decision. The second is a reason to investigate, particularly when critical history is involved.

Consequences of a wrong merge

Describe who could see or act on the combined record. A mistake could expose someone else’s conversation, misassign work, or cause contact with the wrong person. Preferences can become misleading if their original context disappears.

Do not average these dimensions into a reassuring score. One unresolved identity contradiction or unknown critical effect can block an otherwise straightforward merge.

3. Route cases with this filled decision table

The following hypothetical cases form a reusable routing artifact. Replace their details with evidence from each proposal; keep the columns consistent so reviewers can compare decisions.

CaseFour-dimension assessmentRouteRequired evidenceStop condition
A: Repeated intake import; no later activityIdentity: clear shared origin. History: minimal. Recovery: confirmed. Consequences: limited.Single authorised operatorRecord IDs, original submission reference, absence of later activity, retained values, recovery method.Unexplained difference or newly discovered linked activity.
B: Shared phone; different names and contradictory notesIdentity: disputed. History: conflicting. Recovery: does not resolve identity. Consequences: another person’s history could be misattributed.Hold for identity resolutionSource of number, original contradictory context, and a stated question that further evidence must resolve.The shared number remains the only justification; contradiction remains unresolved.
C: Confirmed identity; extensive histories and conflicting preferencesIdentity: resolved. History: complex. Recovery: must be established. Consequences: incorrect work ownership or preference use.Independent second review; hold until handling is clearIdentity confirmation, relationship inventory, preference provenance, owner handoff, recovery limits.Critical history treatment or preference context cannot be determined.
D: Apparent duplicate; linked-activity recovery unknownIdentity: not yet resolved. History: moderate. Recovery: unknown. Consequences: not yet bounded.Hold; reassess after investigationSource records, affected activity, confirmed merge behaviour, recovery limits, consequence assessment.Identity remains unresolved or any critical effect remains unknown.

Case D cannot move directly from “unknown recovery” to approval by a senior person. First establish what the merge will do. If that investigation shows known irreversibility, use the following higher-risk authority rule.

An irreversible merge requires all four conditions:

  1. Resolved identity evidence: The same-person conclusion is supported and known contradictions are resolved.
  2. Understood, bounded consequences: The affected records, relationships, and plausible harms are identified; critical effects are not speculative.
  3. Explicit linked-history plan: The proposal explains attribution, open work, ownership, preferences, and what cannot be restored.
  4. A named independent decision-maker: A person other than the proposer, explicitly authorised by the team to accept irreversible consolidation, records the decision and rationale.

Before using this route, the team must name that decision-maker and a qualified alternate in its operating procedure. If neither is available, hold. A job title alone is not this authorisation. Unknown critical behaviour remains a hold even when an authorised person is willing to accept risk.

4. Give reviewers a decision packet they can challenge

A request reading “Please approve duplicate” encourages rubber-stamping. Provide the identity case, proposed changes, recovery limits, concrete consequences, and requested decision. Reference original evidence rather than unnecessarily copying conversation content into another document.

The following filled packet extends hypothetical Case C. It demonstrates how a reviewer can approve a revised plan rather than merely endorse the proposer’s conclusion.

  • Proposal: Combine fictional contacts E-104 and F-228, retaining E-104 as the surviving record.
  • Identity evidence: A recorded clarification through the team’s existing identity-confirmation process explicitly connects both records to the same person. The reviewer can inspect the original context; no identity contradiction remains.
  • Change map: Retain the current confirmed contact details. Keep each conversation’s original context identifiable. Move responsibility for the open enquiry to its existing owner, Maya; do not interpret consolidation as a new assignment.
  • Preference conflict: E-104 contains an earlier permission note; F-228 contains a later WhatsApp opt-out. The proposal initially described both as a general contact preference. That description is withdrawn. Preserve the dated, channel-specific context and honour the opt-out.
  • Recovery account: For this fictional case, investigation has established that original field values can be restored, but the combined conversation relationships cannot be fully separated. This is known irreversibility, not an assumed undo option.
  • Consequences: Combining the wrong people would misattribute conversations and could expose them to staff handling the other person’s enquiry. Resolved identity evidence and the history plan address the identified risk; irreversible attribution remains an accepted limitation.
  • Proposer: Leo, operations coordinator.
  • Independent decision-maker: Priya, the team’s named authority for irreversible merges, who was not the proposer.
  • Decision: Approve only the revised plan preserving conversation context and the opt-out. Maya will check the resulting record against the change map. If the expected history treatment is unavailable, do not execute; return for review.

The useful intervention is specific: the review changes preference handling and makes the irreversible limitation explicit. It does not claim that reviewer attention alone makes consolidation safe.

5. Make independence useful without creating a bottleneck

Choose reviewers according to the difficult part of the case. A history-heavy proposal needs someone who understands the linked work. A disputed-identity proposal needs someone able to examine contradictions and request appropriate clarification. Seniority alone provides neither skill.

Two people looking at the same matching phone number do not create independent evidence. The reviewer needs access to the underlying context and permission to reject the proposal. Record a reasoned outcome: approve this plan, request specified evidence, or keep the records separate.

Keep well-defined low-risk cases outside the pre-approval queue, then sample them afterward. If sampling reveals overlooked contradictions, reconsider similar recent cases and tighten the boundary that failed. This reduces unnecessary waiting only if operators can recognise escalation triggers.

For urgent work, separate the business need from consolidation. Staff may be able to continue handling the correct existing record while a merge waits. Urgency does not establish that two records represent one person. When the usual reviewer is unavailable, use a qualified alternate or hold—not silent self-approval.

6. Check execution and respond to a mistaken merge

After execution, the assigned checker compares the result with the approved change map: retained values, history attribution, open work, ownership, and preferences. A discrepancy is an operational finding, not something to dismiss because the proposal was approved.

If a mistake is suspected:

  1. Contain further effects: Stop related consolidation and pause affected downstream actions where possible. Alert staff relying on the disputed identity.
  2. Establish the changed state: Identify what changed during and after the merge before attempting repair. Avoid overwriting the evidence needed to reconstruct events.
  3. Recover within verified limits: Use the established restoration method. Where full separation is unavailable, assign a controlled reconstruction plan and record remaining uncertainty.
  4. Diagnose the failed assumption: Distinguish missing identity evidence, overlooked history, execution error, and unexpected software behaviour. Each requires a different correction.

A merge does not itself establish permission to message. Recipient permission and opt-outs still matter; consult the current WhatsApp Business Messaging Policy for precise requirements. Preserve enough context to interpret preferences instead of treating a combined field as fresh permission.

7. Pilot the rules and interpret the findings

Run a small pilot. For example, a hypothetical pilot could assess 20 proposed merges before execution. That batch size is illustrative, not an evidence-based threshold. Choose a manageable batch that includes different case types rather than only obvious duplicates.

Record each proposal’s four-dimension assessment, route, missing evidence, decision, waiting time, and any post-merge discrepancy. Have another reviewer independently route a selection from the recorded evidence, then compare:

  • Routing agreement: Do disagreements reveal missing facts or different interpretations of the authority boundary?
  • Useful intervention: Did review change the plan, uncover a contradiction, or prevent unsupported consolidation?
  • Avoidable delay: Were low-risk cases escalated because operators could not tell what qualified?
  • Recovery readiness: Did a decision rely on an unverified restoration assumption?
  • Execution accuracy: Did the resulting record match the approved plan?

A low escalation rate is not automatically good; a high approval rate does not prove effective review. Repeated missing evidence suggests a packet problem. Repeated routing disagreements suggest unclear authority. Unexpected history changes require investigation of merge behaviour, not merely another approver.

Before adopting this workflow, check relevant capabilities in your workspace and current plan. Alactic’s CRM, Communications, and ACID business-assistant offerings should not be treated as interchangeable feature sets. This process can use an existing controlled review record and does not assume a built-in merge approval or recovery feature.

Contact identityCRM operationsData qualityApproval workflows

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