What it does

Creates a voice workstream and binds it to a phone number so inbound calls can be received and routed. Configures language, hold music, operating hours, recording, and transfer behaviour at the workstream level.

Key facts

• Voice workstreams are push mode only: pull mode unavailable
• A phone number can only be assigned to one workstream at a time
• Anonymous inbound calls supported only on ACS Direct Routing numbers: not on ACS managed numbers
• Multi-language: a single number can serve multiple languages via a language selection menu; add additional languages on the Language page
• Bridged vs non-bridged external transfer:
  • Bridged: Dynamics number shown as caller ID, recording continues through transfer, no DTMF to external, post-call survey triggers
  • Non-bridged: Customer's number shown as caller ID, recording stops at transfer, DTMF supported, no post-call survey; Direct Routing customers need SBC configured to accept REFER-TO
• Operating hours at workstream level override queue-level overflow settings: this conflicts with voicemail configuration (see feature-voice-voicemail.md)

When to use / skip

Essential for any inbound voice deployment. Every inbound number has to bind to a workstream here, no way around it.

Configuration decisions

• Bridged vs non-bridged transfer: Bridged keeps recording rolling and shows your Dynamics number; non-bridged shows the customer's real number and drops recording. Pick based on compliance and callback needs.
• Multi-language: one number vs one per language: Single number with a language IVR menu is cleaner to maintain. Raise this early in discovery; clients often default to separate numbers without thinking it through.
• Post-call survey: Wired in during the Behaviours step of setup, not after. Do this now, not post go-live.

Gotchas

• Workstream-level operating hours kill voicemail. Set at workstream level, and queue-level overflow (which handles voicemail) never fires. If voicemail is in scope, put operating hours at the queue level instead: see feature-voice-voicemail.md.
• Post-call survey lives in Behaviours, not as a separate step. If you need it, link it during channel setup now. Easy to miss otherwise.
• Test all inbound paths before go-live. Voice routing issues are a nightmare to debug remotely. Build a test call plan covering every number, every language, and every operating hours scenario and run it through UAT.

Consultant notes

• The multi-language vs. multiple numbers decision almost always needs to happen during discovery, not design. Clients tend to have a strong opinion once they understand the implications, and retrofitting the wrong choice mid-build wastes time.
• Post-call survey needs to be scoped before configuration starts. It's wired in during the Behaviours step, not added on afterwards: easy to skip in a busy build phase and then have to backtrack.
• Build a test call plan for UAT that explicitly covers every number, every language, and every operating hours scenario. Voice routing issues are difficult to debug remotely and the last thing you want to find in go-live week.

---

Source last updated: 2026-03-10 | Worth revisiting if enhanced voice rolls out to more regions or anonymous inbound stops being Direct Routing only