How to test internet speed

A useful internet speed test is a fair measurement of the network you actually use, taken with as little extra traffic as you can reasonably pause. It is not a hunt for the highest screenshot. It is not a comparison against a stranger’s result on a different continent. SpeedMetric’s internet speed tester runs in the browser: download, upload and ping, typically in about eight to twelve seconds, with no account. The quality of the number depends almost entirely on how you set the scene.

If you tap Start while a console is downloading a game, a TV is streaming 4K, and a laptop is uploading raw photos, you will get a true picture of a busy house — and a terrible picture of the plan you pay for. Both pictures have a use. Decide which one you want before you start. For diagnosing “why is this call ugly,” a busy house is relevant. For checking whether the line can reach the advertised megabits, you want a quiet path.

Before you tap Start

Pause cloud backups, game updates and other video streams. Stay on the Wi-Fi or Ethernet you meant to measure. On a phone, turn off cellular if you wanted home Wi-Fi; otherwise you are measuring the carrier, not the router. If a VPN is on, decide whether you want the VPN path or the ISP path. A VPN often changes the result, which is expected, not a bug in the tester.

Stand where the problem happens, then run a second test near the router. Those two locations answer more questions than ten repeats on the sofa. If the computer has an Ethernet port, or you can borrow a USB adapter, a wired run is the cleanest view of the last mile. You do not need laboratory conditions. You need one variable at a time.

What the result is actually saying

Download is how fast data reaches you. Upload is how fast you send. Ping is delay, in milliseconds. Together they answer whether this is a good internet speed for what you do, better than any one headline. Units are megabits per second. If a download window shows megabytes per second, you are looking at a different label. Mbps versus MB/s is the conversion so a “slow download” is not just a factor of eight.

A high download with a weak upload is a common asymmetric plan. It is fine for watching video and frustrating for two-way calls. A healthy download with a ping over 100 ms will still feel late in a lobby. Jitter — how much ping jumps — is what makes a call sound like it is underwater even when the megabits look generous. Read all three numbers, not only the big one in the middle of the gauge.

Make the test comparable

Testers disagree with each other because they use different servers, durations and methods. That is normal. Compare like with like: same device, same room, same time of day, same tester. SpeedMetric is a browser measurement of this path. Another site might sit closer to a content network you care about, or farther away. Neither result is “fake.” They are different journeys. For what a tester can and cannot claim, see what an internet speed tester measures.

Time of day belongs in the notes. A quiet 2 a.m. run describes the best case. Dinner-time Wi-Fi describes the case you actually live with. If you are collecting evidence for an ISP, three runs on Ethernet or next to the router, morning and evening, are worth more than a single low result on a phone in the garden. Debugging a slow connection is the same idea: same tester, different hop.

Phones, laptops and the radio you forgot about

An old phone with a tired Wi-Fi chip will lose to a recent laptop in the same room. That is not proof the provider throttled you. Test the device that feels slow, then a second device on the same network. If only one is weak, look at that device’s Wi-Fi, power saving, or VPN before you escalate. If both are weak near the router, the access point or the WAN is a better suspect.

Cellular tests are valid measurements of mobile data. They are invalid measurements of home broadband. The public IP will also change when you switch, which is expected — each network has its own address, as the IP lookup will show. Keep the network name in the note next to the Mbps so you do not mix the two later.

A simple routine you can repeat next week

Quiet the house. Test next to the router. Test where you sit. If you can, test on a cable. Write download, upload, ping, the clock time, and the room. Do that again on a busy evening. You now have a small table that explains most “my internet is slow” threads without opening the router admin page. When the far room is the only weak cell, improving home internet speed starts with placement, a better band, or a mesh node — not a new contract.

Common questions