Live chat for multiple websites and brands: four operating models

Four models

An instance, inbox, project and free workspace are not equivalent

A generous domain allowance can still create substantial operational work. Separate projects may each require configuration, payment, monitoring and access reviews. A shared workspace may reduce switching yet expose too many brands, visitors or settings to the same people.

WebChat offers a concrete Teams-centred pattern: the page or department selects an instance and target channel. The reviewed documentation does not establish consolidated cross-channel reporting, tenant inheritance or every possible domain topology. Treat these as proof-of-concept and contract questions.

Sources: 1, 2, 4, 5, 6

Target architecture

Give every website a controlled route instead of copying scripts without an inventory

• Record brand, language, widget or instance ID, destination channel or inbox, owner and deputy for each domain.
• Separate public brand labels from internal Team names so visitors never see confidential tenant or channel terminology.
• Choose deliberately between one shared channel, a channel per brand, a language channel or temporary campaign routes.
• Grant agent and admin rights at the separation unit and test joiners, movers, leavers and external agency access.
• Define common service metrics even when products aggregate by channel, inbox or project.
• Version the script, channel ID, consent text, CSP, tag-manager rule and rollback for every website.

Sources: 1, 7, 9, 10

Cost and governance

The billing unit determines how cost scales

WebChat describes a channel model, Tidio paid subscriptions per project and Chatwoot plan- and agent-related fair use for inboxes. tawk.to claims no website limit, but branding, AI, hired agents and other add-ons may cost extra. These structures cannot be reduced to one monthly number without a real quote.

Build one TCO line per website plus shared implementation, translation, consent, testing, reporting, training, maintenance and incident-response costs. Confirm that one client or brand can be exported and removed technically and contractually without affecting any other unit.

Sources: 3, 4, 5, 6

Multi-site pilot

Two contrasting websites are enough for a meaningful test

1. 01

   Choose contrast

   Select a primary brand and a second website with another language, department or agent group.

2. 02

   Create inventory

   Record domain, script, ID, destination, owner, rights, hours, data class and rollback.

3. 03

   Test both routes

   Start on both sites, reach the correct channel or inbox and reply all the way to the visitor.

4. 04

   Test negative separation

   Prove that agents, admins and reports for one brand cannot unintentionally expose or change the other.

5. 05

   Simulate failure

   Test absence, wrong ID, disabled script and unavailable destination with alert and fallback.

6. 06

   Reconcile reports

   Reproduce volume, first response, unanswered chats and exports per site and in aggregate.

7. 07

   Scale the quote

   Apply the real model to all domains, countries, projects, agents and forecast use.

8. 08

   Offboard one site

   Remove the second site, revoke access, export/delete data and prove the first remains unchanged.

Sources: 1, 4, 5, 8