One customer, four conversations

A customer messages on WhatsApp, follows up on Instagram, then opens live chat on the website. To them it is one conversation about one issue. To most support setups it is three unrelated threads, handled by three people, none of whom can see the other two.

The visible symptom is a customer repeating themselves. The underlying cause is that each channel arrives through its own integration with its own identifiers, and nothing joins them.

Threading across channels

The hard part is identity. Channels expose different identifiers (a phone number, a platform-scoped user ID, an anonymous web session) and none of them agree.

  • Deterministic matching first. A verified email or phone number is a reliable join and should be used before anything probabilistic.
  • Session linking. An anonymous web chat becomes attributable once the visitor identifies themselves.
  • Confidence, not certainty. A probable match is surfaced as a suggestion an agent can confirm, never merged silently.
  • Reversibility. Any merge can be undone, because a wrong merge shows one customer another customer's history.
Worth saying plainly

We chose to suggest rather than auto-merge on probabilistic matches. A missed link is an inconvenience; an incorrect one is a data protection incident.

Why search was the harder problem

Once threads unify, volume rises and search becomes the primary way anyone finds anything. Naive search across a unified inbox degrades quickly, because the useful query is rarely a bare keyword.

What agents actually need is search that combines free text with structure: this phrase, from this channel, in the last fortnight, on threads still open. Getting that combination fast at volume meant indexing the structured attributes alongside the message text rather than filtering results after retrieval.

Handover without loss

The purpose of a shared inbox is that a conversation can change hands. That only works if context transfers with it: the full cross-channel history, what has already been promised, and which internal notes are visible to the team but not the customer.

That last distinction is worth building carefully. Internal notes and customer-visible replies living in the same thread, clearly separated, is what stops the handover happening over email instead.

Key takeaways

  • Customers experience one conversation; systems usually record several.
  • Identity resolution is the core problem, not channel integration.
  • Suggest probabilistic merges, confirm deterministic ones, keep both reversible.
  • Search must combine text with structured filters to stay useful at volume.
  • Handover works only when full context and internal notes travel with the thread.

Frequently asked questions

What is a unified team inbox?

A unified team inbox brings messages from several channels (WhatsApp, Facebook Messenger, Instagram direct messages and website live chat) into one threaded view, so a team can see a customer's full history regardless of where each message arrived.

How are messages from different channels linked to one customer?

Through identity resolution. Verified identifiers such as an email address or phone number provide reliable deterministic matches; weaker signals are surfaced as suggestions for an agent to confirm rather than merged automatically.

Explore InboxPulse?

Explore InboxPulse
Share this insight