How can you effectively monitor the performance of Microsoft Teams?

The problem is not that IT teams are inattentive. It is that the standard monitoring stack was not designed to answer the specific question network engineers need to resolve: which site, which segment, which device caused this degradation, and when exactly did it start? Without that answer, every Teams incident becomes an exercise in educated guesswork.

What follows is a structured breakdown of why the usual approaches fall short, what the observability angle adds, and how to frame a more defensible monitoring strategy for unified communications.

What CQD actually measures — and where it stops

The Call Quality Dashboard is Microsoft’s native answer to Teams monitoring. It aggregates telemetry from audio, video, and screen-sharing sessions, generating quality scores per stream. For organizations that have just deployed Teams and need a baseline understanding of call health, CQD is a legitimate starting point. The data it surfaces — jitter, packet loss, round-trip time — reflects real network conditions as seen from the endpoint.

The structural limitation is time. CQD operates on a processing delay that sits around 30 minutes. When a user calls the helpdesk at 10:04 a.m. to report a frozen screen-share that lasted six minutes, the CQD data covering that window will not surface until mid-morning at the earliest. By the time an IT analyst opens the dashboard, the incident has disappeared from the live view and must be reconstructed from historical logs — a time-consuming process that slows resolution and frustrates both user and technician.

The second limitation is scope. CQD captures what happens at the endpoint and through Microsoft’s cloud relay. It does not speak to the local network path that traffic traveled before reaching that relay. A call that degraded because of a congested Wi-Fi access point on the third floor of a branch office will register as a poor-quality stream in CQD, but the dashboard will not identify the access point, the floor, or the office as the origin. The trail goes cold at the edge of Microsoft’s visibility boundary.

The broader context reinforces why this matters. The ARCEP‘s 2025 market observatory on electronic communications reports a 40% increase in videoconferencing data traffic between 2022 and 2024, with Teams accounting for 55% of that traffic in France. More traffic means more pressure on infrastructure segments that were not originally dimensioned for sustained real-time audio and video flows — and more opportunities for degradation to emerge at precisely the points CQD cannot see.

A practical illustration: take a mid-size company running Microsoft Teams across five regional offices. Central IT receives two or three tickets per week describing audio dropouts during afternoon calls. CQD shows acceptable quality scores for those streams. The tickets accumulate, no pattern emerges from the centralized view, and the issue is escalated repeatedly without resolution. This is not a failure of diligence. It is a predictable outcome of using a tool that stops at the cloud perimeter when the actual fault lives 200 meters from the user’s desk.

Dedicated approaches to Phenisys Microsoft Teams observability are specifically engineered to address this gap, correlating call quality signals with local network infrastructure data so that site-level patterns become visible — and actionable — in near real time.

DEX tools: a genuine step forward, with a structural ceiling

Digital Employee Experience platforms represent a more holistic evolution in IT monitoring philosophy. Rather than focusing exclusively on network metrics, DEX tools measure the end user’s perceived quality of interactions with business applications — Teams included. They collect data on device performance, application responsiveness, session quality scores, and user sentiment signals, then aggregate that data to give IT a view of the workforce’s digital experience across locations and device types.

For IT managers dealing with escalating ticket volumes, the DEX value proposition is real. These platforms provide context that CQD lacks: whether a degraded Teams call happened on a device running at 95% CPU, whether the affected user was on VPN, whether the incident correlated with a software update that rolled out that morning. That kind of cross-signal visibility materially shortens the list of hypotheses an analyst has to work through.

The ceiling appears when the investigation needs to go deeper than the device and the session layer. DEX tools are not designed to operate at the codec level. When a Teams audio stream degrades, the root cause may sit in the negotiation between codecs at the moment of call setup, or in the behavior of a specific DSCP marking policy on a router that sits between two floors of the same building. DEX platforms report that the experience was poor. They do not report why the stream behaved as it did at the protocol level.

The practical implication for IT teams is straightforward: DEX tools and CQD are complementary, not competitive. Together they cover the user experience layer and the cloud-side telemetry. What neither covers is the territory between the user’s device and Microsoft’s network edge — which, as the next section explains, is precisely where the majority of Teams incidents originate.

The 60% blind spot: local network issues and site-level correlation

Industry analysis of Teams support incidents consistently points to a striking distribution: approximately 60% of call quality degradations trace back to local network infrastructure rather than to endpoint hardware, user behavior, or Microsoft-side service issues. This figure matters because the entire standard monitoring stack — CQD, DEX, and Teams Admin Center combined — is oriented toward the endpoint and the cloud. The local network segment sits in neither category.

Local network problems affecting Teams calls are diverse in nature. Wi-Fi channel congestion at a branch office during peak hours. A router firmware update that quietly changed QoS behavior. A misconfigured switch port that applies the wrong traffic priority to real-time audio packets. VPN overload at a hub site concentrating traffic from multiple remote locations. Each of these scenarios produces symptoms that look identical in CQD: elevated jitter, occasional packet loss, round-trip time spikes. Without site-level data correlation, the network team is handed a symptom list rather than a diagnosis.

Site-level correlation changes the diagnostic process fundamentally. When a monitoring layer can map Teams quality metrics to specific physical locations — a floor, a building, a campus zone — and simultaneously ingest network telemetry from the infrastructure serving those locations, the pattern recognition becomes tractable. A spike in jitter that affects only users on the second floor of a specific office, coinciding with a bandwidth saturation event on the Wi-Fi controller serving that floor, is no longer a mystery. It is a work order.

According to reporting by Le Monde on the pressure facing collaboration platforms, the rapid normalization of remote and hybrid work has placed Microsoft Teams at the center of enterprise communications in 70% of large French companies — a deployment scale that magnifies every infrastructure gap into a user experience problem that lands on IT’s desk.

The observability approach fills this gap by treating the network infrastructure and the Teams call data as two inputs into a single diagnostic model rather than two separate data streams managed by separate teams. Network engineers gain the ability to validate whether a configuration change they applied actually improved call quality for users at a specific site. IT support analysts can close tickets with a documented root cause rather than a tentative hypothesis. Mean time to resolution drops not because the team became more skilled, but because the relevant data was finally in the same room as the question being asked.

Building a monitoring posture that holds under pressure

The question most IT managers face at this point is not whether their current stack has gaps — the architecture makes those gaps structural, not accidental. The practical question is how to build a monitoring posture that remains defensible when a VP of Sales loses audio in the middle of a client call and escalates to the CIO by end of day.

The market trend, as documented across enterprise IT operations, points toward layering rather than replacement. CQD and DEX retain value in their respective lanes. CQD provides historical trend data that is useful for capacity planning and long-cycle quality reviews. DEX platforms give a workforce-wide experience lens that informs device refresh cycles and VPN architecture decisions. The observability layer sits below and between those two, filling the infrastructure visibility gap that neither was built to address.

Practically, that means defining what data each layer owns and what questions it can answer. Network teams need to know: which sites are generating Teams quality incidents, at what time of day, and which infrastructure segment is implicated. That question is unanswerable from CQD alone. The moment network teams have site-level call quality data mapped against their own infrastructure telemetry, the diagnostic conversation changes from “we think it might be the Wi-Fi” to “access point 4B on floor 3 shows a correlation coefficient of 0.87 with poor call quality scores on Tuesday and Wednesday afternoons.”

The shift this audit typically surfaces is a structural one. Teams monitoring has historically been owned by the collaboration team, operating primarily through Microsoft’s native tooling. Network engineers have been brought in reactively, handed a ticket description rather than telemetry. Closing that operational gap — giving network teams proactive visibility into call quality at the infrastructure level — is what moves organizations from reactive troubleshooting toward a monitoring posture that can absorb the pressure of a 40% traffic growth curve without proportional growth in escalation volume.

The monitoring investments that demonstrate the clearest ROI in unified communications environments are those that give the right data to the right team at the right moment — not the investments that add another dashboard that nobody looks at after the first month. The distinction is whether the tooling was designed to answer the questions that actually drive resolution, or whether it was designed to report on what the vendor could already measure.