Microsoft Teams analytics: what the built-in reports will and won’t tell you

Microsoft Teams analytics isn’t one thing. It’s four separate data sources, sitting in three different admin portals, each with its own retention clock and its own set of things it flatly refuses to tell you.
That matters more than it sounds. Most reporting projects go wrong at the point someone promises a board-level dashboard built on data that Microsoft deletes after 28 days. Or they promise per-person numbers from a report that hides names by default, and nobody notices until the first review meeting.
The good news is that the free, built-in reporting is better than most people realise. Usage, activity, devices, apps, call quality, PSTN minutes: it’s all there, and for a lot of organisations it’s genuinely enough.
The useful question isn’t “what can Teams report on”. It’s “which of these four sources answers the question I’m actually being asked, and will the data still be there when I need to show a trend”.
Quick summary
The single biggest constraint in Teams reporting isn’t what gets measured. It’s how long each source keeps it, and the limits are wildly inconsistent.
Design your reporting around the shortest retention window you depend on, not the longest one you’ve read about. If a number needs to exist in 13 months’ time, something of yours has to be storing it.
What Microsoft Teams analytics actually covers
Start with the inventory, because the single most common mistake is assuming one portal holds everything.
Microsoft’s Teams analytics and reports reference lists over twenty distinct reports in the Teams admin centre alone. The ones most organisations end up using are the Teams usage report, the Teams user activity report, the device usage report, the app usage report and the client usage report. Then there’s a whole voice family: PSTN usage for Calling Plans, PSTN usage for Direct Routing, PSTN minute pools and blocked users. There’s a Teams Premium feature usage report if you’ve bought Premium, plus frontline, Virtual Appointments and Walkie Talkie reports if those apply to you.
Availability varies by cloud. The events usage report, the frontline usage report, the SMS notifications usage report and the Premium feature usage report are all public-cloud only, so if you’re in GCC or a sovereign cloud, check the reference table before you promise anything.
So the inventory is broad. The constraints are where it gets interesting.
The four places your Teams data actually lives
These are genuinely different systems, not four views of one database.
1. Usage and activity reports. Microsoft 365 admin centre and Teams admin centre. Aggregate counts of messages, meetings, calls, active teams, active channels, devices and apps. This is your adoption and licence-efficiency data.
2. Call Quality Dashboard (CQD). A separate portal at cqd.teams.cloud.microsoft. Organisation-wide call and meeting quality: good, poor and unclassified media streams, by subnet, building, device and media type. This is your network and audio quality data. Microsoft notes that records are typically available within 30 minutes of a call ending.
3. Call Analytics. Per-user troubleshooting, inside the Teams admin centre. One person, their calls, their devices, their network, broken into individual call legs so you can see which hop went wrong.
4. Microsoft Graph usage reports API. The programmatic route. Same underlying data as the admin centre reports, but you can pull it on a schedule and keep it.
The practical rule: CQD tells you the network in the Birmingham office drops audio streams. Call Analytics tells you Sarah’s headset is the problem. Usage reports tell you nobody in the finance team has posted in a channel for six weeks. Graph is how you keep any of it longer than Microsoft does.
If you only remember one thing, remember that quality data and usage data live in different places and have different retention rules. People conflate them constantly.
Why your reports show no names, and who can change that
This one catches people out in the first meeting.
By default, Microsoft 365 usage reports conceal user-level detail. Microsoft’s own wording is that “all reports hide user information such as usernames, display names, groups, and sites to help companies support local privacy laws”. That covers user IDs and user principal names, SharePoint and OneDrive site IDs and URLs, group names, and team names in the Teams usage report. It applies to the admin centre, to the Teams admin centre, to Power BI and to the Graph reports API, so you can’t code your way around it.
Changing it is a deliberate act. A Global Administrator goes to Settings, then Org Settings, then the Services tab, then Reports, and ticks “Display concealed user, group, and site names in all reports”. It takes a few minutes to apply.
And here’s the part worth saying out loud in your governance documentation: switching that setting on is a logged event in the Microsoft Purview audit log. Microsoft records the decision. That’s a feature, not a trap, but it does mean the choice to de-anonymise usage reporting should be a documented decision with a named owner and a stated purpose, not something an admin flicks on to answer a one-off question.
Our Microsoft Teams governance post covers how to write that sort of decision down so it survives the person who made it.
Worth knowing
If you’re being asked to report on individual productivity, the concealment default is your friend. It forces the conversation about why before the data arrives, rather than after. Teams usage counts are a measure of activity in one tool, not of output, and a message count has never told anyone whether the work was any good. Be careful about letting an adoption dashboard quietly become a performance management instrument, because that is how you lose the trust the data depends on.
How long the data lasts, and why that decides your reporting design
Retention is the constraint that should shape your whole approach, and it’s inconsistent enough to be genuinely confusing.
Here’s what Microsoft documents, as of October 2026:
- Usage and activity reports show trends over the last 7, 30, 90 or 180 days. But, in Microsoft’s words, “if you select a particular day in the report, the table shows data for up to 28 days from the current date”. So the 180-day trend line exists, while the detailed day-level table behind it does not go back nearly that far.
- Call Quality Dashboard keeps call records for 12 months, “except for end user identifiable information (EUII) fields, which are removed after 28 days”. Microsoft states plainly that “for compliance reasons, EUII data is only kept for 28 days”.
- Call Analytics shows a specific user’s organised and attended calls and meetings for the last 30 days.
- Real-Time Analytics covers a scheduled meeting while it’s in progress and for 72 hours after it ends. After that, you’re back in Call Analytics.
Read those together and a clear design rule falls out. You can answer “has call quality in this building got worse over the year?” from CQD, because the aggregate stays for 12 months. You cannot answer “which named users were affected last quarter?”, because the identifiable fields went after 28 days. And you cannot reconstruct a specific day’s detailed usage from four months ago at all.
So if a number has to be defensible in 13 months’ time, you need to be storing it yourself. Which brings us to Graph.
Getting the data out: the Graph route
The Graph usage reports API is how you build anything that outlives Microsoft’s retention windows.
The pattern is straightforward. A call like getTeamsUserActivityUserDetail accepts a period parameter of D7, D30, D90 or D180, requires the Reports.Read.All permission, and returns a 302 Found with a Location header pointing at a short-lived CSV download URL. You follow the redirect, you get the file, you land it somewhere you control.
The CSV is richer than the portal view: message counts split by channel and private chat, call and meeting counts, meetings organised versus attended, ad hoc versus scheduled versus recurring, audio, video and screen share duration in seconds, last activity date, and whether the user is licensed.
That last column is the quiet commercial one. “Is Licensed” plus “Last Activity Date” is a licence reclamation report, and it usually pays for the work of building it. Pair it with a proper leaver process, because dormant licensed accounts are the other half of the same problem. Our Support team’s Microsoft 365 offboarding checklist covers that side.
Two practical cautions. The concealment setting still applies, so if names are hidden in the portal they’re hidden in the CSV. And the download URL expires in minutes, so this needs to be an automated pull, not a person with a browser.
The Power BI template app question, and a conflict in Microsoft’s own sources
This needs flagging honestly, because the sources disagree and anyone building a dashboard right now will hit it.
For years the answer to “how do we get Microsoft 365 usage into Power BI” was the Microsoft 365 Usage Analytics Power BI template app. In May 2026, Microsoft published Message Centre post MC1324288 announcing its retirement: downloads blocked from 1 June 2026, end of support and pipelines stopping from 1 August 2026, no packaged replacement, with admins pointed at the admin centre usage reports and the Graph APIs instead. Tony Redmond’s write-up at Office 365 for IT Pros covers it, and is blunt that “there will be no replacement of the template app as a single packaged solution”.
Then it got murky. The archived copy of MC1324288 shows the post was updated on 1 June 2026 with a summary reading “You can safely disregard this message, and we will be posting a new message with additional information in the near future”, while the body still described the retirement. The announcement expired on 2 June 2026. And as of October 2026, Microsoft Learn’s Enable Microsoft 365 usage analytics page still documents the template app as current, with no retirement or deprecation notice anywhere on it.
So: an announced retirement, a retraction of the announcement, a promised follow-up that doesn’t appear to have landed, and documentation that still treats the app as live. We can’t tell you which is correct, because Microsoft’s own sources don’t agree.
What we can tell you is what to do about it. Don’t start a new reporting project on the template app. Build on Graph, where the contract is stable and documented, and treat the template app as something you migrate off at a time of your choosing rather than at Microsoft’s. If you already depend on it, check now whether your installation is still refreshing, because that’s a more reliable signal than the Message Centre right now.
This is also a reasonable argument for landing usage data somewhere you own. If you’re already heading that way, the architecture questions in our Microsoft Fabric consultancy work apply directly.
What it costs to go beyond the free reports
Everything above is included with your existing Microsoft 365 licences. Nothing to buy.
The paid step up is Viva Insights, and it answers a different question. Usage reports tell you how much Teams is being used. Viva Insights advanced analysis is about collaboration patterns: meeting load, after-hours activity, focus time, network breadth across the organisation.
As of October 2026, Microsoft’s UK pricing page lists Viva Insights at £3.10 per user per month, on an annual commitment, excluding VAT. Two things come at no extra cost with Microsoft 365: personal insights for individuals, and the Microsoft Copilot Dashboard, though the latter needs a Microsoft 365 Copilot licence to be useful.
The constraint to understand before you buy: Viva Insights aggregates. The minimum group size setting has to be at least five, and Microsoft describes it as “the smallest number of people that Viva Insights considers a ‘group'”. So if your question is about a four-person team, Viva Insights is structurally not going to answer it. For a lot of SMEs, that single fact decides the purchase.
Third-party Teams reporting and call analytics products exist too, and they’re the honest answer for contact-centre style requirements: wallboards, queue statistics, agent-level live call activity. The built-in reporting was never designed for that. We’d rather say so than pretend otherwise.
The verdict, if you want one. Start with the built-in admin centre reports, because they’re free and cover more than most people expect. Add a scheduled Graph pull the moment you need history beyond 180 days or a number that has to be defensible later. Add Viva Insights only if the question is genuinely about collaboration patterns and your groups are big enough to clear the aggregation threshold. Look at third-party tooling only for real-time or contact-centre needs. In that order, every time.
When you don’t need any of this
Plenty of organisations are being sold a Teams analytics project they don’t need. Here’s when to walk away.
- You have fewer than about fifty users. You already know who’s using Teams and who isn’t, because you talk to them. A dashboard will tell you something you could have found out over coffee.
- Nobody has said what decision the data will drive. “We should have visibility” isn’t a requirement. If no one can name the action that a red number would trigger, the dashboard will be built, admired once and never opened again.
- The real problem is adoption, not measurement. If you already know usage is low, more precise numbers about how low won’t fix it. Training will. Our Microsoft Teams training for beginners guide is a better first spend than a reporting project.
- The question is really about individual performance. Teams activity counts measure presence in a tool, not contribution. Building a dashboard to answer that question will damage trust faster than it produces insight, and the concealment default exists partly to slow that down.
- You need it once. If this is a one-off licence audit or a single board paper, export the CSVs by hand. Automating a report you’ll run twice is a hobby, not a project.
The organisations that get real value from Teams reporting almost always have a specific, recurring decision attached: licence renewal, an office network investment, a migration cutover, a governance review with a fixed cadence.
Making the numbers mean something
A short word on interpretation, because the data is easy and the reading of it is hard.
Active user counts go up when you launch something and drift down afterwards. That’s normal, and a dip six weeks after rollout is not a crisis. Message counts fall when teams move conversation into meetings, which might be better or worse depending on what they’re doing. A team showing zero activity for 90 days might be dormant, or it might be a project archive that’s doing exactly what it should.
Pair every metric with the decision it informs, and write that pairing down. “Active teams with no activity in 90 days” is useful because it feeds an archiving decision. “Total messages sent” is useful to almost nobody.
That’s the discipline the Microsoft Teams governance consultancy work is built around: deciding what you’ll do with a number before you start collecting it.
Worried you’re about to build a dashboard on data Microsoft is going to delete?
It’s a fair worry, and it’s the single most common reason Teams reporting projects get rebuilt a year later. We’ll map the question you’re being asked against what each source actually keeps, and tell you where you need to store data yourself. If it turns out the built-in reports already cover you, we’ll say so and you’ll have spent nothing.
The bottom line on Microsoft Teams analytics
There’s more reporting in Teams than most people use, and less history in it than most people assume.
The four sources answer four different questions. Usage and activity reports tell you how much Teams is being used, over 7 to 180 days, with names hidden until a Global Administrator decides otherwise and Purview logs the decision. Call Quality Dashboard tells you how the network is performing, keeping aggregate records for 12 months but stripping identifiable fields after 28 days. Call Analytics tells you what happened to one named person in the last 30 days. Graph is how you keep any of it for longer.
Get the retention right and everything else follows. Design around the shortest window you depend on, pull to somewhere you own if a number has to survive, and don’t spend money on Viva Insights until you’ve checked that your groups clear the five-person aggregation floor.
And be honest about the question. Most Teams reporting requests are really governance requests wearing a dashboard costume: someone wants to know whether this thing is under control. That’s a better question, and it’s answerable. If that’s the conversation you’re having, our Microsoft Teams governance consultancy work starts exactly there.
STRAIGHT ANSWERS
Frequently asked questions
How can Microsoft Teams analytics help us understand team performance?
It can tell you how much a team is using Teams, and that's a narrower thing than performance. You'll see active users, active channels, posts, replies, mentions, reactions, meetings organised and last activity date, per team. That's genuinely useful for spotting dormant teams, uneven adoption and licences going to waste. It won't tell you whether the work is any good, and treating message volume as a productivity measure tends to produce more messages rather than more output. Use it to find teams that need help, not to rank people.
What reporting platforms are most useful for tracking user engagement and adoption in Microsoft Teams?
For most organisations, the Microsoft 365 admin centre and Teams admin centre usage reports are the right starting point, because they're included and they cover active users, devices, apps and per-team activity over 7, 30, 90 or 180 days. If you need history beyond that, or a figure you can defend in a year's time, add a scheduled pull from the Microsoft Graph usage reports API into storage you control. Viva Insights is the paid step up, and it's about collaboration patterns rather than adoption counts.
What kinds of admin reports does Microsoft Teams provide for tracking call activity and team performance?
The Teams admin centre carries more than twenty reports. The commonly used ones are the Teams usage report, user activity report, device usage report, app usage report and client usage report. On the voice side there's PSTN usage for both Calling Plans and Direct Routing, PSTN minute pools and a blocked users report. Quality sits elsewhere: Call Quality Dashboard for organisation-wide call quality and Call Analytics for one named user's calls. Some reports are public-cloud only, so check Microsoft's reference table if you're in GCC.
How can we use conversation analytics on Microsoft Teams calls?
Carefully, and probably not with the built-in tooling. Call Analytics and Call Quality Dashboard are about technical call quality: jitter, packet loss, device, network path. They don't analyse what was said. Content-level conversation analysis means either Copilot-based meeting summarisation, which is a per-meeting assistive feature rather than a reporting dataset, or a third-party contact-centre product. Both raise consent and employment-law questions in the UK that are worth settling before any pilot.
Is there a way to see live, real-time Teams reporting?
Partly. Real-Time Analytics in the Teams admin centre shows telemetry for a scheduled meeting while it's in progress and for 72 hours after it ends, which is designed for troubleshooting a meeting that's going badly right now. Call Quality Dashboard is near real time, with records typically available within 30 minutes of a call ending. There's no built-in live wallboard for call volumes or queue statistics. If that's what you need, it's a third-party contact-centre tool.
Why do Teams usage reports show anonymous names instead of real users?
Because that's the default. Microsoft hides usernames, display names, groups and sites in usage reports to help organisations meet local privacy law. A Global Administrator can turn concealment off under Settings, Org Settings, Services, Reports, and the change applies to the admin centre, the Teams admin centre, Power BI and the Graph API alike. Worth knowing before you do: switching it on is recorded in the Microsoft Purview audit log, so make it a documented decision with a named owner rather than an ad hoc change.
How long does Microsoft keep Teams analytics data?
It depends which source, and the inconsistency is the thing to plan around. Usage and activity reports show 7, 30, 90 or 180-day trends, but a specific day's detailed table only goes back up to 28 days. Call Quality Dashboard keeps records for 12 months, minus the end-user identifiable fields, which go after 28 days. Call Analytics covers a user's last 30 days. Real-Time Analytics covers a meeting plus 72 hours. If you need anything longer, you have to store it yourself.
Do we need Viva Insights, or are the free reports enough?
For adoption, licence efficiency and quality troubleshooting, the free reports are usually enough. Viva Insights answers a different question: how people collaborate, meeting load, after-hours patterns, focus time. It's listed at £3.10 per user per month in the UK on an annual commitment, excluding VAT, as of October 2026. The catch for smaller organisations is the aggregation threshold, which must be set to at least five people, so questions about small teams can't be answered by it at all.
Can we get Teams analytics into Power BI?
Yes, but build it on the Microsoft Graph usage reports API rather than the old Microsoft 365 Usage Analytics Power BI template app. Microsoft announced the template app's retirement in May 2026, then updated the announcement to say it could be disregarded, and the Learn documentation still describes the app as current. The position is unclear. The Graph route is documented, stable and gives you data you own, which is where you want to be either way.
What's the difference between Call Quality Dashboard and Call Analytics?
Scope. Call Quality Dashboard is organisation-wide and aggregate: good and poor media streams across subnets, buildings, devices and media types, which is how you find a network problem affecting a whole office. Call Analytics is per-user and forensic: one person's calls and meetings over the last 30 days, broken into individual legs so you can isolate the faulty hop. Access differs too, with Teams communications support specialist and support engineer roles seeing progressively more detail than each other.
KEEP READING
