A multi-site operations company was done patching aging PBX hardware. They wanted Microsoft Teams as the phone system: same numbers, remote staff with the same dialing experience as the office, auto attendants that still felt like “calling the company,” and emergency calling that wouldn’t become a compliance afterthought.
- 4 sitesVoice cutovers completed
- 1 pilot siteBefore headquarters move
- Direct RoutingCertified SBC design
- Legacy PBXRetired after final site
What success meant here: keep local numbers, preserve receptionist workflows, support mobile/remote users in Teams, and cut sites one at a time so a bad Friday at one location didn’t take down the whole company.
The situation
Each site had its own habits: different auto attendant trees, different after-hours behavior, and a front desk that knew exactly which buttons meant “don’t mess this up.” Remote staff already lived in Teams for chat and meetings but still juggled a desk phone culture for external calls.
Leadership’s constraint was simple: voice is unforgiving. Email can wait an hour. Missed customer calls cannot.
Design decisions we locked early
- Direct Routing with a certified SBC rather than forcing a one-size calling plan that didn’t fit number strategy
- Number inventory first, every DID, fax edge case, and department line accounted for before cutover talk
- Emergency calling requirements by site, not as a final-week checkbox
- Call queues and auto attendants rebuilt to match real receptionist behavior, not a default Microsoft template
- Pilot site before HQ so edge cases showed up where the blast radius was smaller
How delivery worked
Discovery that maps people, not just ports
We walked call flows with the people who answer phones. Diagrams matter; so does hearing “when it rings twice, we transfer to warehouse.” That detail becomes queue timeouts and holiday greetings later.
Platform build
SBC, voice routing policies, caller ID behavior, emergency locations, and user voice enablement were built and tested in a controlled environment. User training materials used their actual attendant trees, not generic screenshots.
Pilot site cutover
Inbound, outbound, queues, voicemail, mobile clients, and after-hours routing were validated on a smaller site. Issues found: a couple of mislabeled numbers, one queue overflow path, and headset guidance for people new to softphone life. Fix first, then scale.
Staggered site go-lives
Remaining sites followed a repeatable checklist: freeze window, number porting/coordination steps, smoke tests, front-desk shadow support, and a short hypercare period. Headquarters went last on purpose.
Cutover checklist (condensed)
| Before | Day of | After |
|---|---|---|
| Number map signed off Policies assigned Users trained Emergency locations set |
Port/cut steps Inbound/outbound tests Queue & AA checks Front desk live support |
Quality monitoring Exception cleanup PBX decommission gate Lessons into next site |
Results
- Legacy PBX retired after the final successful site cutover
- Users placed and received calls in Teams using existing business numbers
- Auto attendants and queues matched prior front-desk behavior with fewer hardware dependencies
- Remote staff got the same dialing experience as office users without a second phone system
Voice is unforgiving. The pilot site exposed every edge case before we touched headquarters.
- VP of Operations, multi-site company
When Direct Routing is the right call
Not every org needs Direct Routing. Some are happier on Calling Plans or Operator Connect. This client needed number control, multi-site realities, and SBC flexibility. We chose the path that matched constraints, not the path that looks newest in a feature matrix.
Planning Teams Voice?
We can inventory numbers, design call flows, and run a pilot-first Direct Routing or Operator Connect plan.
Talk to a voice specialist