Teams Meeting Random Instability (200+ Participants) - How to Identify the Source of the Problem

Brown Evie 0 Reputation points
2026-08-26T23:44:39.99+00:00

Hello

We are a large school based in USA and sometimes do large Teams conferences with more than 200 people. Around 200/220. We have an A5 license. We inject a video production into Teams, we create a meeting and not a webinar or Live Event. I'm in charge of the video part. Every month, these conferences go smoothly, and then sometimes, data loss, instability cuts the audio for 1, or 2, 3 seconds maximum. The video may freeze from time to time. AJd we had audio cuts every 5 minutes, then it intensified for 15 minutes and then everything stabilized afterwards. This is a recurring problem. I have access to the Teams admin console. The PC behaves correctly. I am aware that the reasons can be numerous, but since it is systematic, I look for all the leads to check. I look at the logs but would like to know how to target the (numerous) data to be checked as a priority in these logs. We think it is a problem related to Teams rather than our infrastructure, because the data measured does not report network anomalies in our area. In short, we want to study each avenue.

Thank you for your help.

Microsoft Teams | Microsoft Teams for education | Meetings and calls | Audio and video
0 comments

2 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. Kien 895 Reputation points Independent Advisor
    2026-08-27T01:07:25.3266667+00:00

    Dear @Brown Evie,

    To resolve this, Microsoft’s support team will need to step in. A technical support engineer can perform a remote session to investigate the situation verify the backend configurations, and run any necessary synchronization tools to resolve the problem. If the issue requires further attention, they can refer it to a specialized team for deeper analysis.   

      Here’s what your Office 365 Global Admin should do next:       

    1. Submit a Service Request: The Global Admin should create a service request with Microsoft support to get advanced technical assistance and diagnostics. For detailed instructions on how to get support, please refer to Get support - Microsoft 365 admin.     
    2. Find Your Microsoft 365 Admin: If you’re unsure who the Global Admin is within your organization, you can find guidance on locating them here: How do I find my Microsoft 365 admin? - Microsoft Support.    
    3. Call Microsoft Support: For additional assistance, use this link to find the appropriate contact number: Global Customer Service phone numbers - Microsoft Support

    In the meantime, kindly try to test this again in other devices and check if the issue persists. Although current PC behaves correctly, it would be helpful if you could test the issue on another computer with a different configuration.

    Was this answer helpful?

    0 comments

  2. AI answer

    2026-08-26T23:45:06.6+00:00

    For a recurring large-meeting issue like this, the fastest way to isolate the source is to use the Teams admin center meeting investigation view and prioritize the telemetry that most directly maps to the symptoms: packet loss, round-trip time, jitter, freeze metrics, frame rate, and device/encoder failures.

    1. In the Teams admin center, go to Manage users.
    2. Select the organizer or the affected production account.
    3. Open the Meetings & calls tab.
    4. Select the affected conference from Completed meetings or In-progress meetings.
      • For completed meetings, telemetry processing can take 30 minutes to 2 hours after the meeting ends.
      • While a meeting is live, data updates every minute.

    Once inside the meeting details, focus on these sections first:

    1. Issues
      • This is the best starting point because it already groups problems by Audio, Video, Screenshare, and Other.
      • Review the listed participants affected, possible root cause, and recommended actions.
      • For recurring incidents, compare whether the same issue type appears during the same 10-minute windows.
    2. Participant-specific quality For the production/injection endpoint and any users reporting impact, prioritize these metrics: Audio
      • Jitter: high jitter can make audio choppy.
      • Packet loss: direct indicator for audio gaps.
      • Local healed ratio: high values are experienced as choppy audio.
      • Round-trip time: high RTT can cause audio latency and overlap.
      Video / screensharing
      • Packet loss
      • Round-trip time
      • Frame rate: low values can indicate system resource constraints, insufficient bandwidth, or capture-device issues.
      • Normalized freeze duration: directly maps to visible freezing.
      • Harmonic frame rate: low values can indicate jitter, freezes, burst arrivals, or low FPS.
      • Loss recovery attempt rate: repeated recovery attempts usually point to network loss.
      • Source freeze count: useful if the injected feed or screen-share source stops producing frames.
      • Audio/video sync: helps detect timing drift between modalities.
    3. Device and driver indicators Since the PC appears healthy, still verify whether the affected sessions show changes or anomalies in:
      • Microphone driver
      • Speaker/headphone driver
      • Camera or screenshare capture device driver
      • Video encoder / decoder hardware failure A hardware encoder failure can force fallback behavior and degrade performance if CPU headroom is limited.
    4. Use thresholds to triage quickly In real-time analytics, these are strong indicators to prioritize:
      • Application sharing packet loss: good threshold is less than 5%
      • Application sharing RTT: good threshold is less than 500 ms
      • Normalized freeze duration: good threshold is less than 25 ms/min
      • Harmonic frame rate average: good threshold is greater than 0.15
      • Loss recovery attempt rate: good threshold is less than 21
      • Source freeze count: good threshold is less than 75
      • Audio/video sync: good threshold is between -900 ms and 900 ms
    5. Interpret the pattern Based on the symptoms described:
      • Short audio drops plus video freezes usually point first to packet loss, high RTT, or jitter.
      • If the issue intensifies for 10–15 minutes and then clears, check whether the spike is visible in loss recovery attempt rate, packet loss, normalized freeze duration, or harmonic frame rate during that exact window.
      • If network metrics stay clean but freezes still occur, check frame rate, source freeze count, and hardware failure flags on the production endpoint.
    6. For the injected production feed If the conference uses a produced feed into Teams, also check the production side for:
      • sufficient upload bandwidth relative to streaming bitrate
      • CPU load on the encoder or production machine
      • whether the encoder is up to date
      • whether the source audio/video entering the encoder already has issues
      • whether firewall/security controls are affecting required traffic
    7. Use Copilot in meeting investigation if available In the Completed meetings table, selecting Troubleshoot opens Copilot with a prefilled prompt and provides a detailed analysis of the user’s meeting experience. This can help narrow which stream and time window to inspect manually.

    A practical priority order for log review is:

    1. Issues section in the meeting
    2. Production endpoint participant telemetry
    3. Packet loss / RTT / jitter
    4. Normalized freeze duration / loss recovery attempt rate / harmonic frame rate
    5. Frame rate / source freeze count
    6. Driver and hardware encoder failure indicators

    If the same pattern repeats and the admin center data does not clearly isolate the cause, the documented support path is to open a support request from Microsoft 365 Admin Center > Support > Help & Support so the case can be investigated more deeply.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.