← All articles
Technical
12 May 2026 · 7 min read

Preparing your network for Microsoft Teams Voice

When a Teams Voice deployment disappoints, the cause is rarely Teams itself — it is almost always the network underneath it. Voice is unforgiving: a file download that takes a second longer goes unnoticed, but 200 milliseconds of extra latency is audible on every call. Preparing the network before migration is the single highest-value piece of work in a Teams Voice project.

The four things that determine call quality

  • Latency — the round-trip time between your users and Microsoft's network. Keep it low by routing Teams media traffic out to the internet locally rather than backhauling it through a central data centre.
  • Jitter — variation in packet arrival times. Consistent delay is manageable; inconsistent delay causes robotic, choppy audio.
  • Packet loss — even small amounts degrade audio noticeably, because voice packets cannot be retransmitted in time to be useful.
  • Bandwidth — usually the least of the four problems on modern connectivity, but still worth modelling for concurrent call volumes at each site.

Quality of Service is not optional at scale

On a lightly loaded network, voice may sound fine without any traffic prioritisation. The problems appear at the busiest moments — which is exactly when calls matter most. QoS marks Teams voice packets so switches and routers forward them ahead of bulk traffic. Implementing it end-to-end across the LAN, and verifying the markings survive each hop, is standard practice for any deployment beyond a small office.

Firewalls and local breakout

Microsoft publishes the specific IP ranges and ports Teams media uses. The guidance is consistent: allow this traffic directly to Microsoft, bypassing proxy servers and SSL inspection, which add latency and can break call setup. Sites that send all internet traffic through a central proxy should plan local breakout for Teams media as part of the migration.

Do not forget Wi-Fi

A network assessment that only tests wired connections misses where many users actually make calls. Wireless voice needs adequate access point coverage, sensible roaming configuration and prioritisation of voice traffic over the air. Meeting rooms and open plan areas deserve specific attention.

Assess before you migrate

A proper readiness assessment simulates Teams call traffic from each site over several days and measures latency, jitter and loss against Microsoft's targets. It turns "we think the network is fine" into evidence — and it is far cheaper to fix issues found in an assessment than issues found by users in production.

Talk to an enterprise voice architect

If any of this raises questions about your own estate, we are happy to have an initial, no-obligation conversation.

Get in Touch