Keep Separate People Separate When They Share a Business Number
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A receptionist answers the shop phone, a buyer calls back from it, and another buyer sends a message using the same number. If staff attach all three interactions to the first person saved in the CRM, that person’s history becomes a mixture of other people’s requests and decisions.
Keep the number at the organisation level when it serves a desk, team or business rather than one individual. Maintain separate person records for the people who use it. Attribute each interaction to a person only when the available evidence supports that identification; otherwise, retain it against the organisation and shared endpoint with the speaker marked unknown.
Multiple CRM contacts with the same business phone number are not necessarily duplicates. The important distinction is between the endpoint reached and the person who spoke. The worksheet below turns that distinction into a practical recording rule.
1. Decide what the number represents before editing contacts
Start with how the number is used, not where it currently appears in the database. A number copied into a person’s phone field may still belong to a reception desk. Conversely, a number published on a business website might reach one named employee consistently. Its current placement is not proof of ownership.
Use an organisation-level record or equivalent shared-endpoint reference when:
- Different people routinely answer or send messages from it.
- The business describes it as reception, purchasing, the shop or another shared function.
- It stays with the business when the person handling it changes.
- Staff use it to reach whoever is available rather than one named person.
If the use is unclear, label ownership as unconfirmed rather than choosing a person for convenience. Ask a narrow question at the next appropriate interaction: is this the shared purchasing number, or a direct number for the named contact?
A shared number can remain a useful way to reach a person without becoming that person’s unique identifier. Record the relationship as reachable through the shared purchasing line, not as owns this number. That wording prevents future staff from treating the number as conclusive evidence of identity.
2. Separate the organisation, endpoint, person and interaction
A workable design keeps four concepts distinct, even if the CRM stores some of them together:
- Organisation: the business with which you have a relationship.
- Endpoint: the shared phone number and its known purpose.
- Person: a named individual associated with the business.
- Interaction: a particular call, message or exchange, with its own speaker evidence.
The organisation can have several people and several endpoints. One endpoint can serve several people. An interaction may have an identified speaker, multiple speakers or no identified speaker. Do not force those relationships into one contact simply because a phone match returns that record first.
Where your CRM supports these relationships, use them. Where it does not, adopt a consistent account note or interaction record with explicit fields for endpoint, speaker status and evidence. Check the workspace and current plan before assuming a particular relationship field, activity association or review function is available.
Avoid inventing a human called Purchasing Desk just to satisfy a required contact field. If a system requires a placeholder, label it clearly as a shared endpoint and exclude it from person counts and person-based decisions where possible. That is a workaround, not a real individual.
3. Filled worksheet: a fictional shared purchasing line
Fictional worked example: All businesses, people, references and interactions in this worksheet are hypothetical. No customer results are implied. Cedar Lane Fixtures has a reception and purchasing number used by receptionist Mina Shah and buyers Leo Grant and Asha Bell. The worksheet deliberately uses a number label rather than a dialable example.
Part A: endpoint ownership
- Organisation: Cedar Lane Fixtures.
- Endpoint reference: CLF-SHARED-01; full number retained in the organisation’s shared-number field.
- Purpose: reception and purchasing enquiries.
- Ownership decision: organisation-level shared endpoint.
- Evidence: Mina explained that reception and both buyers use the line.
- Person-matching rule: a match on CLF-SHARED-01 identifies the organisation and endpoint only.
- Review trigger: anyone reports that the line has become direct, changed purpose or moved to another business.
Part B: people and known speaker cues
- Mina Shah — receptionist. Reachable through CLF-SHARED-01. Known cue: introduces herself by name and says she is handling reception. Can relay purchasing messages; that role alone does not establish that she made a purchasing decision.
- Leo Grant — buyer, hardware. Reachable through CLF-SHARED-01. Known cue: explicitly identifies himself as Leo Grant during the exchange. A reference to hardware is supporting context, not sufficient identity evidence.
- Asha Bell — buyer, lighting. Reachable through CLF-SHARED-01. Known cue: explicitly identifies herself as Asha Bell during the exchange. A lighting-related request without a name remains unattributed.
- Unknown user of CLF-SHARED-01. Not a separate person record. This is the speaker status used when the evidence does not identify an individual.
Part C: interaction ledger
- Interaction C-01, hypothetical Monday morning call: Mina identifies herself and asks for a revised catalogue to pass to purchasing. Record against the organisation, endpoint and Mina. Note that purchasing is the intended recipient; do not identify either buyer as the speaker.
- Interaction C-02, hypothetical Monday afternoon call: the caller identifies himself as Leo Grant and discusses quote Q-H17. Record Leo as speaker and Q-H17 as the subject reference. Keep the shared endpoint association.
- Interaction C-03, hypothetical Tuesday message: an unnamed sender asks whether the lighting order can arrive earlier. Record against the organisation and endpoint, with speaker unknown. Do not infer Asha from the product category.
- Interaction C-04, hypothetical Wednesday message: Asha identifies herself and refers specifically to the earlier lighting delivery question. Attribute the current message to Asha. Review whether her statement also clearly identifies her as the author of C-03 before updating that earlier interaction.
- Interaction C-05, hypothetical Thursday call: Mina answers, then hands the phone to an unnamed buyer. Record Mina for the opening segment and an unknown buyer for the later segment. If the system supports only one speaker, retain the interaction at organisation level and describe the handover.
Part D: unresolved work
- C-03: confirm whether Asha’s later reference establishes authorship of the earlier message. Reviewer: the staff member managing that order. Until resolved, retain speaker unknown.
- C-05: ask the buyer’s name during the next relevant exchange. Do not guess from the quote or product discussed.
- Existing Mina history: inspect earlier purchasing interactions attached through number matching. Keep explicitly identified Mina interactions; move unsupported person attribution back to the organisation.
To reuse this worksheet, replace the organisation, endpoint reference, people, evidence and unresolved items. Preserve the distinction between a known person’s role and evidence that the person participated in a particular interaction.
4. Set an evidence threshold staff can apply consistently
Not every attribution decision needs a lengthy investigation. It does need an explainable basis. Use three practical statuses: identified, tentative and unknown. Only identified interactions should be treated as established person history.
Identified means the exchange contains a clear identification, such as the speaker giving their name, with no conflicting information. This is an operational attribution, not a security check. Knowing who someone says they are does not establish authority to approve a purchase or access sensitive information.
Tentative means there is a plausible clue but no sufficient identification. A familiar sign-off, a product speciality or continuity with an earlier conversation may help staff ask a better question. It should not silently become a confirmed person link.
Unknown is appropriate when no useful evidence exists or cues conflict. A profile name, saved contact label, voice impression or the identity of the previous speaker is not enough on its own for a shared endpoint.
Capture the reason briefly: name given during call, named introduction in message, or no introduction; shared line. Staff reviewing the record later should be able to understand the decision without reconstructing the whole conversation.
5. Give unknown interactions a useful home
An unidentified speaker should not make an interaction disappear. Store it against the organisation and endpoint, alongside its time, subject, relevant order or quote reference, and unresolved speaker status. That keeps the business history available without contaminating a person’s history.
Keep the subject separate from the speaker. A conversation can concern Asha’s order without Asha participating. Mina can relay Leo’s request without becoming its decision-maker. Record these distinctions in separate fields where available, or in plainly labelled notes.
For example, a hypothetical note could read: Speaker: unknown user of shared purchasing line. Subject: revised delivery request for lighting order. Person mentioned: Asha Bell. Attribution evidence: none. Next clarification: ask who made the request when discussing the revision.
Clarify when it matters to the work, rather than sending extra messages solely to complete the database. If identity affects a commitment, obtain the necessary confirmation before relying on it. If the exchange was only a general opening-hours question, leaving the speaker unknown may be the correct final state.
6. Handle handovers and changing evidence without rewriting history
A shared-line conversation can change speakers midway through. Do not assume the person who answered remained on the line. When a handover is apparent, note the boundary and ask the new speaker’s name if relevant.
Messages can also mix authorship. A receptionist might type a buyer’s request or forward a note. Capture the observed relationship: Mina relayed a request attributed to Leo. That is more accurate than recording Leo as the sender.
Later evidence can resolve an earlier uncertainty, but apply it narrowly. Asha confirming that she wrote Tuesday’s delivery message can support a correction to that message. It does not establish that she authored every earlier lighting enquiry from the shared number.
Retain a short correction note with the previous status, new attribution, reason and reviewer. If a person changes roles or leaves, update their current relationship to the endpoint without moving their historical interactions to a replacement employee. The number’s continuing use does not transfer the former employee’s history.
7. Repair mistaken attribution without merging people
If the shared number has already accumulated under one individual, begin with a bounded repair rather than a mass merge or deletion. Choose the affected endpoint and review interactions where attribution changes a live task or decision first.
- Preserve the current associations. Use an available history record or a review worksheet before changing them.
- Confirm the shared endpoint. Document why it belongs at organisation level.
- Keep distinct people distinct. A shared number is not sufficient grounds to merge Mina, Leo and Asha.
- Reassess each interaction. Retain supported person links; return unsupported ones to organisation-level history with an unknown speaker.
- Review dependent notes and tasks. Correct statements that incorrectly assign a request, preference or decision to a person.
- Record what remains unresolved. Do not replace one unsupported guess with another.
The trade-off is explicit: organisation-level history may require more reading than a neatly consolidated person timeline. But false certainty can mislead the next staff member. A smaller, supported individual history is more useful than a complete-looking one assembled from everyone who used the desk phone.
8. Test the rule on a small review batch
Choose one shared business endpoint and a manageable set of recent interactions. A hypothetical pilot could review the latest twenty interactions; the batch size is a working choice, not a benchmark.
Before review, count interactions linked to a person without recorded identity evidence. After review, count how many have supported person attribution, how many remain unknown, and how many needed correction. A rise in explicitly unknown speakers can indicate more honest recording, not poorer performance.
Ask a second staff member to review a small sample using the same rules. Compare disagreements: did someone treat a product category as identity, miss a handover, or assume the saved number label was the speaker? Refine the worksheet where interpretations differ.
The next action is concrete: complete the ownership and people sections for one shared number, review the selected interactions, and give every unresolved item either a relevant clarification step or an explicit decision to remain unknown. Success is not identifying everyone. It is ensuring that the next person reading the CRM can tell what is known, what is inferred and what has not been established.
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