Measure jitter twice in the browser: at idle and under load. Shows jitter increase, loaded latency, bufferbloat and voice or gaming suitability.
Runs in your browser
The load phase downloads real data from Cloudflare, so it can use a few to a few dozen MB depending on the run.
What this measures
Jitter is the variation between consecutive RTTs. This page measures it twice: once at idle, then again while a download is filling the line.
Both phases use the same 2 s timeout so idle and loaded loss stay comparable.
Why jitter matters more than average ping
A line with 20 ms average latency can still sound awful if each packet arrives at a different time. Voice and video depend on rhythm as much as on speed.
Idle versus loaded jitter
Idle jitter shows the path itself. Loaded jitter shows what happens once your own queue starts filling. The gap between them is often the real cause of “my call breaks only while downloading”.
What to do when load ruins the result
Enable SQM or smart queue management on the router if it supports it.
Compare Wi-Fi and cable. Local wireless contention often shows up as loaded jitter first.
How to use
Run the idle phase first to capture the natural rhythm of the line.
Enable the load phase to see how much the line degrades while traffic is flowing.
Read the diagnosis and the bufferbloat row together. That pair tells you whether the problem is “always bad” or “only bad under load”.
FAQ
Why are there two jitter numbers?
The headline value is the mean absolute successive difference, which is intuitive and comparable to many public tests. The detail row still keeps the loaded and idle latency beside it.
Why can a speed test look fine while this looks bad?
Throughput hides rhythm problems. A download can finish quickly while real-time traffic still suffers from queueing and uneven delivery.