When moving to Teams Phone, failures usually come from the PSTN connection, not the client: Microsoft rejects the SBC’s certificate, outbound calls fall onto the wrong route, or emergency calling goes live without ever being tested. The Teams client easily becomes a corporate phone; how it connects to the public telephone network (PSTN) is a separate architectural decision. That decision determines who runs which component, how number porting is done and how flexible the coexistence period with the existing PBX will be. Below we compare the three connectivity models and give the technical requirements of Direct Routing and the order of checks in the field.
What Teams Phone is made of
Teams Phone is not a single product; it is five parts working together. Leave one out and calls won’t connect:
- Teams Phone license: Enables call control, hold, transfer, voicemail and PBX features for the user. It is included in some Enterprise plans and bought as an add-on in others. Since the scope changes from time to time, assess the Teams Phone need separately when choosing a Microsoft 365 license.
- PSTN connectivity: Calling Plan, Operator Connect or Direct Routing.
- Number: An E.164 phone number assigned to the user or a resource account.
- Voice applications: Auto attendants and call queues; they work with resource accounts.
- Devices: Teams-certified desk phones, conference devices and analog endpoints behind the SBC.
PSTN connectivity options compared
| Criterion | Calling Plan | Operator Connect | Direct Routing |
|---|---|---|---|
| Operator | Microsoft | An operator in the program | Any SIP trunk provider |
| Who runs the SBC | Microsoft | Operator | The organization or a partner |
| Setup | From the admin center | Choosing an operator in the admin center | SBC, certificate, PowerShell |
| PBX integration | None | Limited | Flexible |
| Analog devices | Not supported | Depends on the operator | Through an SBC/ATA |
| Availability | Certain countries | The operator’s coverage | Technically broad |
Which countries and which operators offer Calling Plan and Operator Connect varies; check Microsoft’s current country and operator lists before deciding. In projects in Türkiye, Direct Routing with a local operator’s SIP trunk is a common option. Treat this not as a rule but as a decision made after the operator options have been confirmed.
Decision criteria
- If analog endpoints such as a PBX, door intercom, fax or elevator line will remain for a long time, Direct Routing gives flexibility.
- If you want telephony entirely as a service and don’t want to run an SBC, choose an Operator Connect operator if one serves your country.
- If you have many branches and local calling must continue during a WAN outage, consider Direct Routing with a Survivable Branch Appliance (SBA).
- Models can be mixed: Calling Plan in one country and Direct Routing in another can run in the same tenant.
Direct Routing architecture: SBC, SIP trunk and certificates
In Direct Routing, the organization’s Session Border Controller (SBC) connects with one leg to the operator’s SIP trunk and with the other to Microsoft’s SIP proxy. Voice, signaling and number translation between these two legs are the SBC’s responsibility, and half of all failures are found here. Requirements:
- Certified SBC: A hardware or virtual SBC from the list Microsoft has certified for Direct Routing must be used.
- FQDN and domain: The SBC’s FQDN (for example
sbc1.example.com) must belong to a domain verified in the tenant;*.onmicrosoft.comcan’t be used. According to Microsoft’s documentation, at least one licensed user must exist in that domain for it to become active. - Public certificate: A certificate that includes the SBC’s FQDN, issued by a root CA Microsoft trusts.
- Public IP: A public IP address on the SBC, directly or via 1:1 NAT.
- Signaling: SIP over TLS; Microsoft’s side uses port 5061 at
sip.pstnhub.microsoft.com,sip2.pstnhub.microsoft.comandsip3.pstnhub.microsoft.com. - Media: SRTP. UDP ports 3478–3481 and 49152–53247 on the Microsoft media processor side; the media port range on the SBC side is defined on the SBC and opened on the firewall accordingly.
Write firewall rules based on the current IP ranges and ports Microsoft publishes; the ranges can change. If the SBC shows as “not connected” in the admin center, the first thing to check is the TLS handshake: is the certificate chain incomplete, does the FQDN match the name on the certificate, is 5061 really open from outside? If the certificate chain is incomplete, the SBC won’t connect.
When media bypass is enabled, media flows directly between the internal Teams client and the SBC, and Microsoft’s media processors aren’t involved. Latency drops, but clients must be able to reach the SBC’s media ports. If an internal ACL forgets this, the call connects but there’s no audio.
Configuration example
SBC-side settings are vendor-specific (vendors such as AudioCodes and Ribbon publish configuration guides for Teams). The Teams side is done with the MicrosoftTeams PowerShell module:
Connect-MicrosoftTeams
# Add the SBC to the tenant
New-CsOnlinePSTNGateway -Fqdn sbc1.example.com -SipSignalingPort 5061 `
-MaxConcurrentSessions 60 -Enabled $true
# PSTN usage, route and voice routing policy
Set-CsOnlinePstnUsage -Identity Global -Usage @{Add="TR-All"}
New-CsOnlineVoiceRoute -Identity "TR-Route" -NumberPattern "^\+90\d{10}$" `
-OnlinePstnGatewayList sbc1.example.com -Priority 1 -OnlinePstnUsages "TR-All"
New-CsOnlineVoiceRoutingPolicy -Identity "TR-Policy" -OnlinePstnUsages "TR-All"
# Assign a number and the policy to a user
Set-CsPhoneNumberAssignment -Identity [email protected] `
-PhoneNumber "+902125550101" -PhoneNumberType DirectRouting
Grant-CsOnlineVoiceRoutingPolicy -Identity [email protected] -PolicyName "TR-Policy"
The route pattern in this example only covers ten-digit numbers starting with +90. Defining separate routes and separate PSTN usages for international and emergency numbers makes it possible to restrict at policy level who can call where. For users’ short-dialing habits (such as dialing 0 for an outside line), normalization rules are written in a tenant dial plan in Teams. If normalization is wrong, the call matches no route and the user only sees “call failed.”
The coexistence period with the existing PBX
Within the existing systems infrastructure and PBX setup, it is rarely possible to move all users in one night. The method that works in the field is placing the SBC between the operator and the PBX:
- The operator trunk connects to the SBC; the PBX connects behind the SBC via a SIP trunk.
- The SBC routes incoming calls to Teams or the PBX according to the number dialed. When a user is moved, their number is switched to the Teams side in the SBC routing table.
- Calls between PBX extensions and Teams users also flow through the SBC; a normalization rule is defined in the dial plan so that short extension dialing keeps working.
- Users are moved in department groups; the call center, reception and analog endpoints move last, with a separate test plan.
If Operator Connect or Calling Plan is chosen, numbers need to be ported to the new operator or to Microsoft. The process depends on the operator and local regulation; clarify with the operator in advance whether partial porting is possible. If a block will be ported in one go, the Teams side for all users in that block must be ready before the porting date. A half-prepared block means a half-working PBX.
Auto attendants, call queues and the number plan
The greeting menu and hunt groups on the PBX are rebuilt in Teams with auto attendants and call queues. Each uses a resource account; the free “Teams Phone Resource Account” license is assigned to the resource account and, if an external number is needed, the number is given to that account.
- Auto attendant: Business-hours and after-hours menus, holiday calendars, dial-by-extension and speech recognition options.
- Call queue: Takes agents from a Teams team or group; the distribution method (attendant routing that rings all agents at once, serial, round robin, longest idle) and overflow and timeout destinations are defined.
- Number plan: Prepare a table mapping the extensions on the PBX to DID numbers. In Teams every user is defined with a full E.164 number; for extensions without a DID, an extension (
;ext=) can be used.
A common gap is that the PBX menus were never documented. Before the migration, map the existing call flows one by one with test calls; then no branch is missing on the Teams side, and nobody has to ask in production “where did after-hours callers end up?”
Network readiness and voice quality verification
Teams audio is real-time; latency, jitter and packet loss matter more than bandwidth. Microsoft’s network readiness documentation targets one-way latency under 50 ms, round-trip latency under 100 ms and jitter under 30 ms from the client to the edge of Microsoft’s network. Confirm the values in the current documentation at the start of the project.
- Local internet breakout: Carrying Teams media through a central proxy or VPN tunnel adds latency. Direct breakout and VPN split tunneling are recommended for endpoints in the Microsoft 365 “Optimize” category.
- QoS: The port ranges and DSCP values Microsoft recommends mark audio on UDP 50000–50019 as EF (46), video on 50020–50039 as AF41 (34) and screen sharing on 50040–50059 as AF21 (18). Marking is done in Windows with Group Policy QoS policies; preserving the marking across switches and the WAN is the network side’s job. Queue design on the campus network is covered in our QoS on the corporate LAN guide.
- Measurement: The Microsoft Teams Network Assessment Tool tests media quality and connectivity from branch networks to Microsoft;
NetworkAssessmentTool.exe /connectivitycheckverifies access to the required endpoints. - Monitoring: After going live, upload subnet and building information to the Call Quality Dashboard; poor-quality calls are then grouped by location.
A packet marked at the client is reset when it passes through an untrusted switch port along the way. If the marking is lost on the way, there is no QoS.
Migration checklist
- PSTN model confirmed against country and operator availability.
- Licenses assigned for user and resource accounts.
- SBC certificate, FQDN, public IP and firewall rules ready; the SBC is enabled in the Teams admin center and the TLS connection shows as “active.”
- Number inventory, DID mapping and porting date finalized with the operator.
- Dial plan normalization rules, voice routing policies and international calling restrictions tested.
- Emergency call routing verified with a test call.
- Auto attendant and call queue flows tested with business-hours, after-hours and holiday scenarios.
- Network assessment done at branches and QoS markings verified.
- The pilot group has used the system for at least one working week and feedback has been collected.
The cloud side of a Teams Phone migration is intertwined with Microsoft 365 operations and the network side with infrastructure design. To plan both together, see our Microsoft 365 and cloud services and infrastructure consulting pages. We covered how to plan Microsoft 365 services against outages in our article on continuity in Microsoft 365 and cloud infrastructure. Microsoft’s Plan Direct Routing document should be the reference for the current Direct Routing requirements.
Frequently asked questions
Do you always need to buy an SBC for Teams Phone?
No. With Calling Plan or Operator Connect, the SBC is run by Microsoft or the operator. You only need your own SBC when Direct Routing is chosen, when analog device integration is needed or when long-term coexistence with the PBX is planned.
Which certificate should be used for Direct Routing?
You need a public certificate that includes the SBC’s FQDN, issued by a root certificate authority Microsoft trusts. Microsoft’s SIP proxy doesn’t accept a certificate issued by an internal CA. The FQDN’s domain must be verified in the tenant.
Can the existing PBX and Teams Phone run at the same time?
Yes. The SBC connects to both the operator line and the PBX and routes calls to Teams or the PBX based on number ranges. Users are moved in groups; analog devices and the call center move on their own schedule.
Where should you look first for voice quality problems?
Per-user call analytics in the Teams admin center and the Call Quality Dashboard are the first step. Then check the client’s network, the wireless connection, media flowing over VPN and whether QoS markings are preserved along the path.