M365 Teams | Admin Center: Meeting quality insights

The Teams Admin Center is getting insights that flag meeting quality issues early. They identify hotspots, likely causes, affected users, and recommended actions. The announcement is available in the Message Center under MC1523077.

You don’t need to set thresholds. Detection uses machine learning on historical data and decides for itself what’s anomalous. This saves configuration, but there’s a trade-off: you can’t see the logic behind what’s flagged as an anomaly. So, cross-check the results against your own data.

The insights update daily and are based on your tenant's meeting data. SIP phones, PSTN, and peer-to-peer calls are excluded. So they won’t help you with telephony.

You can find the Insights in the Teams Admin Center under Meetings > Meetings. Access is available to Global Administrators, Teams Administrators, Teams Communications Administrators, Teams Communications Support Engineers, and Teams Communications Support Specialists.

The global rollout begins mid-November 2026 and ends early December. In GCC, it also starts mid-November and ends mid-December. No activation is required.

The Insights do not replace the existing tools. According to Microsoft Learn, Call Analytics provides per-user details on devices, network, and connection for individual calls and meetings from the last 30 days. The Call Quality Dashboard (CQD) displays quality organization-wide. Call data is typically available there within 30 minutes after the call ends.

The new insights shorten the search for the problem. For root causes at the network level, you still need building and endpoint data in CQD.

If you haven't uploaded building and endpoint data yet, now is the time. The upload is done in the Teams Admin Center. Only then will CQD display location-based reports and distinguish whether a problem affects a single user or an entire group.

Insights provide a no-configuration starting point but are not a replacement for Call Analytics and CQD. From mid-November, check whether the suggestions align with your metrics, and clarify role assignments beforehand.