The three deployment models
Microsoft Calling Plans
Native Microsoft PSTN service sold directly by Microsoft, bundled into Microsoft 365 licensing. Simplest path. Highest per-user cost in most scenarios. Availability varies by country and is shallower than it appears once you look at the specific regions where your users actually work.
Operator Connect
PSTN service from Microsoft-approved telecom partners (Verizon, Vonage, Pure IP, BT, and others), provisioned through the Teams admin center. Middle ground: partner telecom pricing, partner telecom SLAs, streamlined provisioning. Broader international coverage than Microsoft's native Calling Plans.
Direct Routing
Self-managed or partner-managed SIP trunk to Teams via a Session Border Controller (SBC). Most control, most flexibility, highest setup complexity. Requires SBC hardware or cloud-SBC service and the operational capacity to run it. Often the lowest per-user cost at scale.
These are not a good/better/best ladder. They are three different architectures with different fit criteria.
The decision tree
Our team uses a three-question decision tree with every Teams Voice client:
This isn't a substitute for a real evaluation, but it gets most clients to the right model in a 30-minute conversation instead of a multi-month investigation.
Honest TCO math
Teams Voice TCO comparisons are the single most miscalculated number in UC evaluations we see. The honest model for a 250-user mid-market organization:
The "savings" of Direct Routing relative to Calling Plans can easily be offset by operational overhead if the organization isn't staffed to run it. Above 1,000 users, Direct Routing is almost always the right answer on pure economics.
The SBC question
If you're going Direct Routing, the Session Border Controller is the architectural decision that most affects your operations for years. Two paths:
Hardware SBC
Physical appliance (AudioCodes, Ribbon/Sonus, Oracle/Acme Packet). One-time capital cost, ongoing maintenance, rack-and-stack somewhere. Best for: organizations with existing on-prem telecom infrastructure, specific latency or compliance requirements, predictable call volumes.
Cloud SBC
SBC-as-a-service from providers like AudioCodes Live Platform, Ribbon Cloud, or Microsoft's own offerings. Subscription cost, zero infrastructure, faster deploy. Best for: organizations without existing telecom infrastructure, moderate call volumes, faster-to-production priority.
We recommend cloud SBC for most clients under 1,000 users. Above that threshold, the economics of hardware-plus-maintenance start penciling more favorably.
The adoption problem
Technical deployment is the easy part. User adoption is where most Teams Voice deployments under-deliver versus their business cases. Three patterns we see consistently:
- Partial migration syndrome, some users on Teams Voice, some on legacy PBX, receptionists handling both. Persists for 18+ months because no one champions the final cutover.
- The "phones in drawers" problem, users still have desk phones for calls they could make on Teams. Hard deadline for deskphone decom is the only effective treatment.
- Missed training on telephony features, call queues, auto-attendants, hunt groups, compliance recording are routinely under-trained. A 2-week focused enablement push post go-live moves ROI meaningfully.
Numbers and porting
Number porting is the other predictable friction point. Three rules from 500+ deployments:
- 1Start the porting process 6 weeks before go-live. Carriers are inconsistent; some are 2 weeks, some are 8. Six minimum.
- 2Do a soft-cutover test the week before. Route a subset of calls through the new system with the old as fallback. You will find issues before go-live, which costs much less.
- 3Have the previous carrier on notice, not on short fuse. Keeping the old carrier active for a week post-cutover is cheap insurance.
