You see who attended a Teams meeting through the meeting's attendance report or by opening the participant roster during and after the meeting. In Teams, organizers and co-organizers can open the meeting chat or calendar event and select Attendance to view join and leave timestamps, duration, and role for each participant. The exact path depends on your license, role, and tenant settings.

How Teams turns connections into attendance rows

Every attendance row starts as a signaling event, not a click on a camera or microphone. When a participant connects, Teams negotiates a media session through Microsoft 365 and records join and leave events against that meeting's identity. The attendance report is a derived view of those events, not a separate attendance database.

This distinction explains most confusion. Audio-only dial-in participants, room systems, and anonymous users can appear as rows even though they never opened the desktop app. Conversely, a person who watches a live stream of a Teams webinar may not appear as an attendee because they never joined the interactive meeting session. The report answers "who connected to the meeting," not "who paid attention."

The mechanism has a timing consequence. Join and leave times are computed from these connection events, so a participant who drops and reconnects may generate multiple intervals or an aggregated duration. Duration reflects time connected to the session, not engagement with content. A participant who joins, mutes, and walks away accrues the same connection time as an active listener.

One deeper layer: the attendance report is assembled from meeting telemetry that Microsoft retains and exposes through the Teams client and, in some configurations, through Graph or PowerShell. Access is governed by role, licensing, and tenant policy, which is why two people in the same meeting can see different amounts of data. Plainly: attendance is a permissions-gated transcript of connection events, not a universal record.

Quantified context: what the report actually contains

A standard attendance report lists each participant with a name or identifier, a role, a join time, a leave time, and a duration. It also shows the meeting's start and end times and the organizer. For meetings with many participants, the list can be long enough that the client paginates or requires a download to review fully.

Timing precision is the detail most people overestimate. Connection events carry timestamps, but the displayed times are rounded and can lag the true connection by seconds to a minute or more depending on network conditions and client state. Treat join and leave times as close approximations, not forensic markers. If you need precise arrival times for compliance, say so to your IT administrator, because the client view is not an audit-grade record.

Retention and export behavior also matter. The client report can typically be downloaded as a file for offline review, which is useful when you need to share attendance with someone who lacks meeting access. Webinar and large-meeting attendance data may require the organizer to open a separate attendance view. Where your organization licenses broader meeting analytics, IT can pull attendance through administrative tooling rather than the client.

Scope matters for numbers. Attendance reflects the participants in that specific meeting instance. A recurring series produces separate attendance per occurrence. A channel meeting may attribute attendance differently from a private scheduled meeting. State clearly which meeting instance you are reviewing before comparing counts.

Edge cases that break naive attendance counts

The attendance list is accurate for connection events but misleading for interpretation. Several participant types and behaviors inflate, deflate, or distort the count.

Dial-in and room-system participants can join audio without a named account, producing identifier rows that are hard to map to people. Anonymous or unauthenticated attendees may appear with generic labels, especially in meetings configured to allow external participation without sign-in. Reconnects and device switches can create duplicate intervals for one person, so raw row counts exceed headcount. Breakout rooms split the session; participants may appear in breakout attendance separately from the main room. Live-stream viewers in webinars and town halls may not register as interactive attendees at all. Recording and transcription do not equal attendance; a participant can attend without recording, and a recording can exist with a minimal live audience.

The counterintuitive professional caveat: a higher attendance count is not better evidence of participation. If your goal is verifying who received information, pair attendance data with a poll, a reaction, an assignment, or a transcript review. Attendance proves a connection; it does not prove comprehension. Professionals who report attendance as engagement are quietly overstating what the data supports.

Misconceptions about visibility and tracking

The most common misconception is that anyone in the meeting can see attendance. In practice, visibility is asymmetric. Organizers and co-organizers generally have the attendance view; ordinary attendees usually do not. This asymmetry is intentional, because attendance data can reveal schedules and participation patterns.

A second misconception is that attendance is real-time for everyone. The roster in the meeting shows who is present now, while the post-meeting report shows connection intervals. These are different surfaces with different refresh behavior. The roster updates as people join and leave; the report solidifies after the meeting.

A third misconception is that deleting or hiding a participant erases their attendance. Attendance records follow meeting and tenant retention policy. If your organization retains meeting data, removing a name from a chat does not remove the connection event.

The fourth misconception is that attendance equals presence in a physical sense. Distributed work means many participants join from phones, cars, and shared devices. A person can be connected through a device while physically elsewhere. The report measures account and device connection, not location or attention.

Attendance surfaces compared

Surface Who can see it What it shows Best use Key limit
In-meeting roster Organizer and participants Current participants and roles Managing a live session Not a historical record
Post-meeting attendance report Organizer, co-organizers Join/leave times, duration, role Verifying who connected Rounded times; role-gated
Downloaded attendance file Whoever receives the export Same data offline Sharing and archiving Snapshot; can go stale
Admin meeting analytics IT administrators Aggregate and per-meeting data Compliance and reporting Requires licensing and policy
Chat or transcript review Meeting participants Messages and captions Evidence of participation Does not prove attendance

Practical actions to get reliable attendance data

Start by deciding what question you are answering. "Who attended" and "who participated" require different evidence. If you need connection proof, use the attendance report. If you need engagement proof, add a poll or an explicit acknowledgment request during the meeting.

Concrete steps for organizers:

Confirm your role before the meeting. Only organizers and co-organizers reliably get the full attendance view; ask the organizer to add you as a co-organizer if needed. Open the meeting's attendance view after the meeting ends, and download the file if you need to share or archive it. Reconcile identifiers with your roster. Match dial-in and room-system rows to people, and flag anonymous rows as unverified. De-duplicate reconnect intervals before counting headcount. Report unique participants and total connection time separately. Set a threshold for what counts as attendance. A common practical rule is a minimum connected duration, such as several minutes, to exclude accidental joins; state the threshold you used. For recurring meetings, review each occurrence. Aggregate counts across a series hide per-session drop-off.

Reconciling thresholds explicitly prevents disputes. If one report uses "any connection" and another uses "at least a few minutes connected," the counts will differ. Pick one definition, document it, and apply it consistently. US organizations setting internal reporting standards should align attendance definitions with their retention and privacy policies rather than inventing ad hoc rules.

Limitations and what attendance cannot tell you

Attendance data is bounded by licensing, role, and tenant configuration. Some organizations restrict export or limit how long meeting data is retained. Client views may round timestamps, and administrative views may differ in scope. None of this means the data is wrong; it means the report is a tool with a defined purpose.

It also cannot measure attention, comprehension, or intent. A participant connected for the full hour may have been entirely elsewhere. A participant who joined late and stayed briefly may have absorbed the key decision. If decisions depend on attendance, supplement the report with explicit acknowledgments, polls, or follow-up confirmations.

Finally, privacy obligations apply. Attendance data can reveal working hours and participation patterns, and in some jurisdictions it is personal data subject to notice and retention rules. Consult your organization's privacy or legal guidance before using attendance records for performance evaluation or discipline; this article is general educational information, not legal advice, and it does not promise any outcome from using attendance data.