Call Quality, Monitoring And Interpreting How-To
- Introduction
- Configure and Start Monitoring
- Before You Begin
- Monitor a User for a Selected Period
- Let Users Monitor Their Own Calls
- Start Monitoring an Active Call
- Read and Troubleshoot Reports
- Open a Completed-Call Report
- Read the Summary First
- Identify the Affected Audio Direction
- Understand the Measurements
- Worked Examples
- When Results Are Missing or Inconclusive
- Export Call Audio Metrics
- See Also
Introduction
Use Monitor Connection Quality to collect call audio statistics and identify which connection and audio direction may be affected by a network problem. The report separates the caller and callee, so an administrator can investigate the relevant user connection or SIP trunk.
Enable monitoring before reproducing an issue, or start it while the problem call is still in progress. Opening an old call in the call log displays statistics already collected for that call; it does not collect missing data retrospectively.
Configure and Start Monitoring
Before You Begin
- Use a current 3CX app or the Web Client connected to your PBX.
- To monitor another user or inspect administrative reports, sign in with the appropriate role and department access. Use a System Owner or System Admin account when investigating across the system.
- Identify the affected user, the approximate call time, the call parties, and the symptom. Record who heard the problem and whether it affected one or both directions.
If the option is missing, check the account permissions and department scope. On the Team page, scroll to the bottom of the user's three-dot menu before assuming Monitor Connection Quality is unavailable. You do not need the affected user's password or access to their desk phone to enable monitoring from Team.
Monitor a User for a Selected Period
Use this method for an intermittent issue so that calls made during the selected period can be investigated.
- In the Web Client, open Team and locate the affected user.
- Open the user's three-dot menu and select Monitor Connection Quality.
- Choose 1 day, 3 days, 5 days or 7 days in the For Period list.
- Ask the user to make or receive calls that reproduce the issue, and note the time of each affected call.
- After each test call ends, open its call quality report as described below.
Repeat this for each affected user. Monitoring one extension is not a system-wide monitoring setting. Choose a period that covers the investigation, and review or disable monitoring from the same menu when the investigation is finished.
Let Users Monitor Their Own Calls
A user can enable monitoring for their own extension in the same style; in the Web Client, click the profile avatar at the top right and select Monitor Connection Quality.
In the Windows Softphone, open the extension avatar menu; in the Android and iOS apps, open the main menu. Select Monitor Connection Quality.
After a monitored call ends, the user can review the result in call history. For the detailed report, open Calls in the Web Client and select Show call quality report for that call.
Start Monitoring an Active Call
While a problem is active, open Panel, locate the connected call under All Calls and open its context menu. Select Monitor Connection Quality, where available for your account. Keep the call connected long enough to reproduce the symptom, then end it and inspect the report.
If monitoring is started partway through a call, it will only cover the monitored portion. For a repeatable test, enable monitoring for the user before making the next call.
A user can initiate Monitor Connection Quality during an active call directly from the soft clients:
- Softphone: Right click on the caller name, and click on Monitor Connection Quality.
- Mobile Apps: Tap on the call duration label for your call, and tap OK on the confirmation dialog.
Read and Troubleshoot Reports
Open a Completed-Call Report
- As an administrator, open Admin > Reports > Call Log and find the call by its time and participants.
- Open the call's actions menu and choose Show call quality report. Users can open the same action for their own calls from Calls in the Web Client.
- Confirm the caller, callee and reported duration match the call you are investigating.
- Read the summary, then compare both audio directions for each participant.
A disabled report action means a report is not available for that entry. Ensure you are viewing the relevant call segment, especially after a transfer or queue journey.
Read the Summary First
The summary identifies detected network problems and the affected participant or direction. "No problems detected" means the collected statistics did not identify a problem. "Overall score" summarizes the available stream quality; it is not a guarantee that every part of the conversation sounded clear.
Stream ratings are Excellent, High, Medium, Low, Poor or Unknown. Medium, Low and Poor indicate a stream worth investigating. Unknown means the statistics needed to rate that stream were not available; it does not mean the stream was good or bad.
An "Incomplete" summary means there is insufficient information for a complete conclusion. Check the available participant and direction details before deciding what to test next. An overall rating can look good while an unreported direction remains Unknown.
Identify the Affected Audio Direction
Read the direction relative to 3CX, separately for the caller and callee. A call has two participants, and each participant has an audio path to 3CX and an audio path from 3CX.
- "Audio to 3CX" under the caller: audio travelling from the caller towards the PBX. Problems here can affect what the callee hears.
- "Audio from 3CX" under the caller: audio travelling from the PBX towards the caller. Problems here can affect what the caller hears.
- "Audio to 3CX" under the callee: audio travelling from the callee towards the PBX. Problems here can affect what the caller hears.
- "Audio from 3CX" under the callee: audio travelling from the PBX towards the callee. Problems here can affect what the callee hears.
The participant shown for an external call represents the path between 3CX and the external side of the call. A problem on this leg is a reason to investigate the trunk/provider path; it does not, by itself, identify a fault at the remote caller's device or prove which network equipment caused it.
Understand the Measurements
- Score / MOS: an estimate of voice quality derived from network statistics. Compare the individual stream ratings before relying on the overall score.
- Packet Loss %: the proportion of audio packets reported lost in that direction. Loss can cause gaps, broken speech, or robotic audio. Compare a problem call with a clear call from the same endpoint.
- Packet Count: the number of packets represented in the report. Interpret it alongside the monitored duration; a short sample provides less evidence than a representative test.
- Jitter (ms): variation in audio packet arrival timing. Higher variation can contribute to uneven audio when the receiving device cannot compensate for it.
- RoundTripTime (ms): the time for traffic to travel between the participant and PBX and back. This is a round-trip measurement, not a one-way mouth-to-ear delay. Endpoint reporting limitations can make it unavailable or estimated.
- Packet Bursts / Average: burst count and average burst length reported for the stream. Consider these alongside loss, jitter and the user symptom rather than treating the count alone as proof of a fault.
- Codec, Endpoint / UserAgent and Duration: identify the audio format, device or app, and the collected sample length. Compare like-for-like calls where possible.
- Local Address / Global Address / Tunnel: identify the addresses visible to 3CX for that participant. A Web Client's Local Address can show 127.0.0.1 for the WebRTC proxy; this is not the user's device address. A Global Address may also be shown. A tunnel address may represent an SBC or app connection. Remove public addresses before sharing report screenshots externally.
If Mode is shown, Proxy means 3CX relayed the audio packets; Transcoding means 3CX processed the audio. These describe the audio path and are not quality warnings on their own.
Use the ratings and measurements together. Call averages can hide a brief interruption, and the report does not diagnose a microphone, headset or speaker fault. If someone heard poor audio despite good network statistics, keep investigating the endpoint and the other parts of the call path.
Worked Examples
The first example uses the supplied demonstration report, with its public IP address replaced by a documentation example address. The remaining examples use invented extensions and illustrative values. These examples explain how to read the report; they do not define fixed MOS thresholds.
Example 1: Packet Loss Towards the Callee
In the demonstration call, April May (1000) is the caller and Additions Support (1013) is the callee. The summary identifies "Network problems from PBX to 1013 Additions Support", and the Overall Score is Poor.
Under the callee's "Audio from 3CX", the Score is Poor and Packet Loss is 31.58%. The other three audio directions are Excellent with 0% reported packet loss. This points to audio travelling from the PBX towards the callee, so the callee may hear gaps or broken speech from the caller.
Start by investigating the path from the PBX to the callee. The callee's jitter is only 0.3 ms and RoundTripTime is 3 ms, so low delay and jitter do not cancel out the packet-loss finding. Both directions labelled "Audio to 3CX" are Excellent; the data does not point to an upload problem from either participant.
Both endpoints are identified as "3CX WebRTC proxy", with Local Address 127.0.0.1. Use the participant names and direction to identify the affected connection. The displayed Mode, Transcoding, describes audio processing and is not itself the fault.
Example 2: The Caller Sounds Broken to the Callee
The Caller (201) shows Poor for "Audio to 3CX" with 6% packet loss, while the other reported directions show High or Excellent. The Callee (202) reports gaps in the caller's speech. Start by investigating caller 201's path towards the PBX: their Wi-Fi or wired connection, possible uplink congestion and internet connection. Repeat the test on a wired connection or an alternative network and compare the same direction.
Example 3: One Direction Is Unknown
Caller 201 has an Excellent reported stream, but Callee 202 shows Unknown for "Audio from 3CX". The good rating does not clear the unreported direction. Repeat a monitored test and check whether the endpoint or provider supplies the required RTCP quality statistics.
When Results Are Missing or Inconclusive
- Monitoring was not active: enable it before the next call and reproduce the issue. Historical calls cannot be measured afterwards.
- Media bypassed the PBX: direct audio paths can prevent 3CX from collecting the required statistics. An administrator should review the audio path if repeated tests produce no usable report.
- An endpoint did not report statistics: missing RTCP reports can leave a direction Unknown. Compare with another supported endpoint or app.
- The call was transferred or routed through a queue: inspect the relevant participant and call segment. Do not assume one report describes every stage of the call.
- The problem was brief or outside the monitored portion: enable monitoring before the next attempt and record the time and symptom precisely.
Change one test condition at a time, such as Wi-Fi versus wired networking or the endpoint used. Compare the affected direction before and after the change. If several users show problems at the same time, investigate shared network or PBX resources; if only external calls are affected, compare internal calls and the trunk path.
Export Call Audio Metrics
Call quality data is exportable through 3CX Data Connectors for external reporting and historical analysis. In Admin > Integrations > Data Connectors, configure a supported database destination and include Call Audio Metrics in the data selected for transfer. Exporting only CDRs or system metrics does not export call quality data.
Exported metrics can be used in tools such as Grafana or Power BI to compare quality over time. Monitoring still needs to collect the call statistics first. Data Connector availability depends on the installed update and license, separately from the monitoring feature; check the Data Connector requirements displayed by your PBX.
See Also
- Call Reporting
- Connecting 3CX to a Standalone PostgreSQL & Grafana
- Grafana Reporting Configuration
- Power BI Reports and Dashboards
- Access Roles in the Web Client
Last Updated
This document was last updated on 23 October 2026
