A single customer view does not come from automatically merging every message with a similar name. It comes from carefully linking channel identities, keeping source and certainty visible and, when in doubt, leaving profiles separate until an employee can confirm the relationship.
That may sound cautious, but an incorrect link is more serious than a temporarily duplicated profile. If an email, phone call, chat or review is attached to the wrong person, an employee may see someone else's order information or act in the wrong file. A safe omnichannel approach therefore prioritises demonstrable identity over an apparently complete customer view.
One person has several channel identities
A customer may use a personal email address, later call from a business number and use a different display name in chat. Conversely, family members may share an email address, a phone number may be reassigned and several contacts may represent one organisation.
Record not just “the customer”, but also the individual identities through which someone appears in a channel:
- normalised email address;
- phone or chat number in a consistent format;
- portal or online store account ID;
- organisation and contact person;
- source, verification status and last use.
These details provide indicators. Not every combination is automatically sufficient to link profiles.
Work with levels of certainty
| Signal | Certainty | Appropriate action |
|---|---|---|
| Logged-in customer adds a new channel themselves | High, provided account checks are correct | Link identity under control |
| Order email address and verified contact match | Relatively high | Link within agreed process rules |
| Name and postcode look similar | Layer | Do not merge automatically |
| Same surname and shared phone number | Uncertain | Conflict for human review |
| Customer mentions an order number in free text | Insufficient for sensitive information | Check ownership further |
Use exact, normalised keys for automatic recognition. Fuzzy matching may suggest candidates, but must not merge customer histories independently.
A conservative merging process
- Normalise: put email addresses, phone numbers and channel IDs into a fixed format.
- Find exact matches: handle them according to the source's verification status.
- Create a separate profile for a new identity: prevent similarity from immediately producing a link.
- Show possible connections: give employees relevant indicators without unnecessarily mixing the two profiles' data.
- Verify: use a reliable channel or a verified account or order relationship.
- Merge with an audit trail: record who linked them, when and on what basis.
- Support recovery: safely undo an incorrect link without losing messages.
The design rule is simple: separate profiles create extra searching; an incorrect merge can cause a privacy and operational incident.
Show a relevant timeline, not an unlimited archive
Once identities are correctly connected, a shared timeline can help. An employee might see that a customer first asked by email and later supplied more information by phone. This stops the customer from having to start again.
But “everything in one place” does not mean every old message should appear by default for each new case. Select by purpose:
- messages about the same order, application or file;
- recent arrangements affecting current handling;
- open actions and previous decisions;
- channel preferences determining the response.
Limit the timeline's length and sensitivity. Employees should be able to open additional history when needed and permitted, but receive only case-relevant information by default.
Retrieve current order data only when needed
A customer profile need not contain a permanent copy of every order. For a service question, the application can retrieve targeted context from the leading source after identity and ownership are verified.
This has three advantages:
- the employee sees a more current status;
- less operational data is duplicated in long-term storage;
- a fact's source stays recognisable.
Yindle Connect can combine relevant data from the online store, ERP and logistics. Yindle Desk can then show only the context belonging to the specific service case.
Treat channel rules as part of the process
Each channel has its own limitations. Chat may have a limited response window, email often contains attachments, and a review is public but may provide insufficient identity for personal order information. Record for each channel:
- how identity is established;
- which information may and may not be shared through it;
- how consent or preference is recorded;
- when the conversation moves to a safer channel;
- how employees see the complete relevant history;
- which retention period and deletion route apply.
A public review may prompt a service case, but must not automatically link to a customer profile based on a name alone.
Design conflicts as normal work
Identity conflicts are not unusual system errors. They are part of real customer data. Create a clear workflow for uncertain cases:
- block the display of sensitive combined context;
- show employees which characteristics conflict;
- let them choose a verification step appropriate to the risk;
- record the outcome and source used;
- correct linked conversations and actions after an earlier incorrect merge.
At most, let AI flag candidates or missing fields. The final link must rely on fixed, verifiable rules or human confirmation.
Checklist for one safe customer view
- Are channel identities stored separately from the main profile?
- Is it visible which identity is verified and by which source?
- Are similar names never used as the sole evidence?
- Can an employee assess conflicts?
- Can an incorrect merge be reversed and audited?
- Does the timeline show only relevant history by default?
- Are current orders retrieved selectively and tied to verified ownership?
- Does every channel have suitable rules for information and responses?
- Does deletion work across profiles, identities and channels?
Also read how to automate online store customer service step by step without losing context or human oversight.
Reliable identity before a complete profile
A good customer view gives employees coherence without mixing different people or organisations based on assumptions. It combines verified identities, a focused timeline and current process information. When uncertain, separation remains the safe default.
Want to organise several channels around one service process? Discuss identity and context architecture with Yindle before automatically merging existing customer data.