Guide

How to accurately test your Wi-Fi speed

And why the number is different every time you look.

A single speed test is a snapshot, not a verdict. Variation is normal. A pattern is the useful part. One fast heat does not prove the house is fine, and one slow heat does not prove the broadband company owes you a refund. Run the test the same way, a few times, in more than one spot. Then read the pattern.

Published 20 October 2026 by Upcott Studios.

Cartoon racers at full speed on a night stadium track, the way a fast wifi heat looks on Lag Chasers — A fast heat. Nitro is a rumour, the crowd is awake, and the number is allowed to look good. It is still only this device, this minute, this path.

What a speed test actually measures

It does not measure “the internet.” It measures the path from the device in your hand to one test server, right now. On Lag Chasers that path is scored four ways: download in megabits per second, upload in megabits per second, ping in milliseconds, and jitter, which is how much those ping samples disagree. The cartoons are the display. The bytes are real. How the measurement works is the short version of the math. Run a heat when you want the number.

A phone on wifi, a laptop on Ethernet, and a console in another room are three different tests. A VPN, a full cloud backup, or twelve tabs of video are also part of the result, whether you meant them to be or not.

Why the result jumps around

Wifi is a shared radio in a noisy room. These are the usual reasons two tests, five minutes apart, do not match:

Interference. Neighbours, microwaves, baby monitors, and a second router on the same channel all shout over your signal.
Congestion. Evening is when everyone else in the house, and on the street, is online. The pipe and the wifi airtime get shared.
Server distance. A test to a nearby edge is not the same as a test that detours across the country. Compare tests on the same tool.
The device. An old phone, a hot laptop, or a weak wifi chip will cap the score below what the line can do.
Background traffic. Updates, backups, and someone else’s 4K stream are using the same connection you are trying to measure.

The same cartoon racers moving slowly: a walking cheetah, a sitting axolotl, a yawning frog, and a raccoon who tripped — A slow heat. Same crew, worse path. One of these does not mean the line is broken. A pile of them, in the same spot, might.

The test worth trusting

Do it in this order. Skip a step and you will argue about the wrong thing.

Ethernet first. Plug a computer into the router with a cable. Pause backups and big downloads. This is the closest you can get to the speed the line is actually delivering.
Wifi next to the router. Same device if you can, or the phone you actually use. Stand in the same room as the router. The gap between this and Ethernet is the normal wifi tax.
Wifi in the problem spot. The bedroom, the garden office, the corner where calls die. This is the number that matters for that room. If only this one is bad, the line is probably fine and the house layout is the problem.

Three-panel cartoon: a laptop on an ethernet cable, a phone beside the router, and a phone in a far room with a weak wifi signal — Cable at the router, wifi at the router, wifi where it actually hurts. Three tests. One story.

How many tests, and when

Run at least three heats in each spot and keep the middle one, not the hero and not the disaster. Then repeat the set at three times of day: morning, evening, and late at night when the street has gone quiet. A line that is fine at midnight and miserable at 8pm is a congestion story. A line that is miserable in every spot, including Ethernet, is a line story.

Write the four numbers down. Download, upload, ping, jitter. The ranks are a joke about the download number. The notes you keep are the actual record.

A worked example

This is a teaching table for a typical 500 Mbps home plan in a two-storey house, not a lab certificate and not your result. The pattern is the point. Ethernet sets the ceiling. Wifi beside the router gives some of that back. The far room falls off a cliff. Evening pushes it further.

Spot	Download	Upload	Ping	Read it as
Ethernet at the router	480 Mbps	42 Mbps	9 ms	Closest to the plan
Wifi next to the router	310 Mbps	38 Mbps	12 ms	Normal wifi tax
Wifi, back bedroom	28 Mbps	11 Mbps	34 ms	Walls and distance
Same bedroom, 8pm	9 Mbps	4 Mbps	61 ms	Congestion on top

Normal wobble, or time to call

Normal: wifi a step below Ethernet, the far room worse than the hallway, evenings slower than mornings, and any two tests within shouting distance of each other. A 400 Mbps result and a 460 Mbps result on a cable are the same story.

Call the provider when Ethernet, in the same spot, at a quiet hour, is far below the plan on several tests. Half the advertised download, over and over, with nothing else using the line, is a fair complaint. Bring the times and the four numbers. Do not bring one screenshot from a phone in the kitchen.

Do not call them first when only one room is bad. That is a placement problem: the router is in a cupboard, the walls are thick, or the room is one floor and a fridge away. A cable into that room, a mesh unit, or a powerline kit fixes the house. The line was never the villain. The gear shelf lists the usual fixes.

Then run it again. One heat on Lag Chasers, written down, is a snapshot. Three spots, three times of day, is a picture you can argue from. If the cable is fast and the phone is not, the next piece is Wi-Fi speed versus internet speed: which link in the chain is actually the slow one.