Network speed test

How it works

You start the test. The tool sends and receives a burst of data to a server and times the burst. Download speed is how quickly that data arrives at your device. Upload speed is how quickly your device can send data out. Ping, or latency, is the round-trip delay for a small request, shown in milliseconds.

Throughput is usually shown in megabits per second (Mbps). Latency is not a speed; it is a delay. A line can move a lot of megabits and still feel sluggish on a call if the delay is high. Read all three numbers. Ignoring ping because the download looks large is how people “pass” a test and then hate a video meeting.

The server you reach is part of the result. A nearby server on a quiet path will often look better than a distant one. That does not mean your provider changed in ten seconds. It means the path changed. A fair comparison uses the same kind of test, on the same device, as close to the same conditions as you can keep.

The test consumes real traffic. It has to: you cannot time a transfer without transferring. On a metered mobile plan, that burst counts. On a shared household line, that burst also counts against whatever else is happening: a backup, a console update, someone else’s stream.

How to make the snapshot honest

If you want the speed of the line into the building, use a wired connection and pause other uses of that line. If you want the speed of the laptop on the sofa, stay on Wi-Fi and accept that walls, distance, and other radios are in the measurement. Mixing those two questions is the usual fight at the kitchen table.

Phone tests and computer tests will disagree when one is on cellular and the other is on the home access point. Say which network you measured. “The network” is not one object. Cellular, guest Wi-Fi, and the main home SSID are different networks that can share a roof.

Note whether you are on Wi-Fi, Ethernet, or cellular. Stop large transfers on that same network before you run the test. Run it twice. Keep both results. Compare like with like: same device, same network, similar time of day.

A single surprising number is a clue, not a court case. Run again. If both runs are far below what you pay for on a wired, quiet line, you have something to take to the provider. If only the sofa is slow, you have a radio problem, not necessarily a line problem.

A speed test measures a moment, not a contract. Run it twice before you argue with the number. Do not stack three testers and average them as if they were the same instrument. Different testers pick different servers and different methods. Pick one, learn its three numbers, and use that one when you compare today with last week.