When a user says "calls are bad,..." pulse helps you answer them.

Microsoft Teams call quality within Pulse

Every voice engineer knows the message. A user pings their it support: "Teams has been bad this week." That gets send to the service provider, a ticket gets opened, and then everyone waits... because working out why one person's calls sound rough has always meant pulling call quality exports, lining them up against session border controller logs, and hoping the pattern jumps out of a spreadsheet before the user gives up and just uses their phone.

The frustrating part is that the answer is usually small. It's almost never the network or Teams; it's one specific path. Let take our figurative person Alex Johnson. Through late April every one of Alex's calls left over the wired office network and scored Excellent: MOSMean Opinion Score, a perceived voice quality metric (1-5) which is  around 4.4, jitter near 1.3 ms, almost no loss. Boring, in the way you want it to be.

Then the bad calls started in early May. Same person, same Teams, same carrier, but on the days Alex worked from home the MOS fell into the 2s. By mid-May most calls were coming over the home Wi-Fi behind one public IP, 203.0.113.x, and every one of them scored Poor: roughly 42 ms jitter, 7.9% packet loss, 208 ms round-trip. The office calls that still happened stayed pristine, so it clearly wasn't Teams. It was that one link.

We wanted to make that obvious without the spreadsheet digging, and not with another dashboard of tenant-wide averages. A tenant average tells you everything is fine while one person quietly suffers. The question engineers actually ask is per-user, so the tool should be too.

That's what the new Call quality tab in Pulse is. Open any Teams user under Identities and it's there: their calls, their devices, their networks, their numbers. No setup if Teams Analytics is already on the service. One day of history is always included; the full 30 days comes with pulse enterprise.

The whole answer on one screen: open the user, and the summary, the devices, and the calls are already there.

The page mirrors how you'd actually debug it. You start with the obvious question: is this person really having a problem, or do they just remember the bad ones? For Alex the 30-day window settles it. 200 calls. 104 Poor, all from 203.0.113.x. 89 Excellent, all from the wired office. 7 with no telemetry. Blended, that comes out to MOS 3.15 and reads "Poor," a single number that tells you almost nothing and that a tenant-wide chart would bury.

From there it narrows. Each device-and-network combination gets its own card with its own score. Alex's laptop, LAPTOP-7F3A2B, splits almost evenly: one wired location averaging Excellent, one Wi-Fi location averaging Poor. That split is invisible in an aggregate. Here it's two clicks.

One laptop, two locations. The wired office is healthy; the home Wi-Fi is the entire problem.

When you need the raw detail (the specific call, the jitter spike, the codec, the media path) the per-call table is right underneath, filterable, with a Details view on every row.

 Down to a single call: MOS 2.1, 41.6 ms jitter, 7.8% loss, over a 40 Mbps home Wi-Fi link at 203.0.113.45.

The plumbing behind it is what callerconnect does. Pulse has graph subscriptions per tenant, device fingerprints normalized so one laptop stays one card. None of that matters to the person opening the tab. They open it and the answer is already there.

For Alex the fix is right there on the screen. Every poor call traces to one egress IP, a roughly 40 Mbps home uplink, while every other location is Excellent. Move that site to a wired connection, or get the home link sorted. You don't touch a Teams policy or open a carrier ticket. We built this because our own support team kept hitting that exact wall: a customer would describe a problem, we'd believe them, and then we'd lose an hour assembling the evidence to confirm what they already knew. Now it's on screen before the call even starts.

It's live now for every Teams user in Pulse. If you've got a list of users you've been meaning to look into, this is a good week to open it.