
Jitter is how much your connection's delay changes from one moment to the next. Packet loss is the share of data that never arrives. In a call, high jitter makes voices sound robotic and loss makes words and syllables drop out, even on a fast connection. Microsoft gives under 30 ms of jitter and under 2% packet loss as typical values for Teams audio. Zoom recommends 40 ms of jitter or less and 2% loss or less. You can measure both in a browser, from the Windows command line, or in the meeting app's own statistics.
The guide to calls that lag on fast internet explains why speed is not the whole story, and the pre-call internet check covers when to test. This guide covers what the two numbers mean and how to measure them.
Jitter and packet loss in a call
Audio and video travel in small packets, many per second. If every packet took the same 40 ms, the call would simply run 40 ms late, and nobody would notice. Jitter is the unevenness: one packet takes 20 ms, the next 70 ms. The meeting app holds packets in a buffer to smooth this out. Microsoft says Teams can adapt to some jitter through buffering, and you only notice it once jitter exceeds what the buffer can absorb.
Packet loss is worse, because a lost packet is gone and the app has to cover the gap. Microsoft's call quality guide describes what each problem sounds like:
| Problem | What it is | What people hear and see |
|---|---|---|
| Jitter | Packets arrive at uneven intervals | A robotic-sounding voice |
| Packet loss | Packets never arrive | Missing syllables and words, choppy video and screen sharing |
| High round-trip time | Packets take a long time to arrive | A walkie-talkie effect, with people talking over each other |
What counts as good
Meeting apps publish their own reference values. They are targets for the app's own call traffic; use them as a scale when you read other tests:
| Source | Jitter | Packet loss | Round-trip time |
|---|---|---|---|
| Teams call health, audio (Microsoft) | Under 30 ms | Under 2% | Under 200 ms |
| Zoom meeting statistics (Zoom) | 40 ms or less | 2% or less | Not given as one figure |
A single reading above these values does not mean the call will fail. A reading that stays high over several minutes, or one that rises whenever someone else starts using the connection, is worth fixing before the call.
How jitter is calculated
Tools measure jitter in slightly different ways, so compare results from the same tool. Cloudflare's speed test, for example, calculates jitter as the average difference between consecutive latency measurements. With five measurements of 20, 24, 19, 45 and 21 ms, the differences are 4, 5, 26 and 24 ms, so the jitter is about 15 ms. The average latency of 26 ms hides that one measurement took about twice as long as the others.
Three ways to test before a call
| Method | Jitter | Packet loss | What it tests |
|---|---|---|---|
| Cloudflare speed test in a browser | Yes, idle and under load | Yes | Your route to the nearest Cloudflare data centre |
ping and pathping in Windows | Not directly | Yes, including where it starts | The route to whatever address you choose |
| The meeting app's statistics | Yes | Yes | The call itself |
1. A browser test that reports both
Open speed.cloudflare.com on the PC you will join from. Cloudflare explains that its test measures jitter both while the connection is idle and while it is busy downloading and uploading, and measures packet loss by relaying packets through a WebRTC TURN server. Watch the gap between idle and loaded results. If jitter jumps once the connection is busy, a backup, cloud sync or another device's upload may be filling the line, and a call would suffer the same way.
2. Windows command line: ping and pathping
Open Command Prompt or Terminal and run:
ping -n 100 1.1.1.1
By default ping sends four requests; -n 100 sends a hundred, which takes a minute or two. The summary line shows how many were lost and the percentage. The minimum and maximum times show how far apart the slowest and fastest replies were.
To see where loss begins, run:
pathping 1.1.1.1
pathping sends 100 requests to every router on the route and shows the loss at each one. It takes a few minutes. Read it carefully: Microsoft's own example shows routers that drop requests addressed to them but still forward everything else. Loss that appears at one router and does not continue to the end of the route does not matter. Loss that starts at a router and continues all the way to the destination is real.
- Loss from the first hop, which is your own router, points to Wi-Fi or your home network.
- Loss that starts further along points to your provider, or beyond it.
3. The meeting app's own numbers
This is the only method that measures the call itself. In Teams, open More actions โ Settings โ Call health during a call; Microsoft says the values update every 15 seconds. In the Zoom desktop app, open Settings โ Statistics. A test call is enough. The guide to Teams test calls and Zoom test meetings shows where to start one.
For Teams, Microsoft also offers the free Teams Network Assessment Tool. It sends packets to the nearest Teams relay and back, then reports loss, jitter and round-trip time.
What to change when the numbers are high
| What you see | Likely cause | What to try |
|---|---|---|
| Jitter is fine at idle but high under load | Something else is filling your connection | Pause cloud sync, backups and large downloads, then test again |
| Jitter or loss is high on Wi-Fi, fine on a cable | The wireless link | Use a cable, or move closer to the router and test again |
| Loss starts at your own router | Your home network | Restart the router and compare cable with Wi-Fi |
| Loss starts further along and continues to the end | Your provider's network, or beyond it | Save the pathping output and send it to your provider |
| Numbers are high only with a VPN connected | The VPN route | Follow the VPN call lag guide |
What CallReady measures
CallReady checks the connection as part of its pre-call check from the Windows tray. It measures latency, jitter and approximate download and upload throughput to Cloudflare, then shows them next to the camera and microphone results. It does not measure packet loss. If people tell you words are dropping out, use the Cloudflare test, pathping or the meeting app's statistics for that.
CallReady measures your route to Cloudflare, not the media path to a Teams or Zoom server. A good result means your own connection is steady. For anything specific to Teams or Zoom, use the app's statistics in a test call.
Frequently asked questions
What is a good jitter for video calls?
Is jitter or packet loss worse for a call?
How do I check packet loss on Windows?
Does CallReady measure packet loss?
Check latency and jitter with CallReady before the call, and use the meeting app's statistics to confirm packet loss.
Get CallReady