How fast is your connection, really?

A fixed amount measures something different on every line: a fibre connection is done in half a second – too short to average out the ramp-up – while a slow line takes two minutes. A fixed time measures the same everywhere: everyone gets a result after seven seconds, and for everyone it rests on an equally long window.

Yes, for latency above all. Latency grows with distance; from the US expect 100 ms and more regardless of your line. Download and upload are less affected on good connections, but a long path with packet loss can reduce them. For a line in Germany, Austria, Switzerland or neighbouring countries the result is representative; from further away, treat latency as a measure of distance, not of your provider.

Four usual reasons, in order: Wi-Fi instead of cable (the most common by far), an old router, other devices using the line at the same time, and the test running on a device that cannot process the data fast enough. Repeat the test with a cable and nothing else running before you call your provider.

The time a packet needs to the server and back. It decides how 'snappy' things feel – remote desktop, voice calls, online games – more than bandwidth does. Below about 30 ms is excellent for a wired line, above 100 ms noticeable; what matters most is that it stays stable rather than jumping around.

No. The test sends random data to our server and reads random data back; nothing about you or your result is kept. The measurement runs entirely between your browser and the test endpoint.