Almost everyone has run a speed test. Far fewer people have run one in a way that produces a meaningful result, and fewer still know what to do with the numbers that come back. A speed test taken over Wi-Fi, at eleven in the morning, on a laptop three rooms from the router, tells you almost nothing about the connection you are paying for.
This guide explains how to test properly, what each number means, and how to use the results to work out whether the problem is your provider, your equipment or your expectations.
How to run a test that means something
Use a cable
Connect a computer to the router with an ethernet cable if you possibly can. Wi-Fi introduces its own limits and its own variability, and testing over it measures your wireless network rather than your internet connection. If you cannot use a cable, stand next to the router and use the 5 GHz band, and treat the result as approximate.
Stop everything else
Pause downloads, streaming, cloud backups and anything else using the connection, on every device in the house. A test run while a console is downloading a game will produce a number that is accurate and useless.
Test more than once, at different times
A single test is a snapshot. Run one in the morning, one in the early evening and one late at night. The evening figure is the one that matters, because that is when networks are busiest and when you are most likely to be using it. The gap between the quiet result and the busy result is often more informative than either number by itself.
Use more than one testing service
Different test servers produce different results. Use at least two independent services, and if your provider offers its own test, treat it as a useful comparison rather than the definitive answer, since it usually measures to a server inside the provider network rather than across the wider internet.
What the numbers mean
Download speed
How quickly data arrives, in megabits per second. This is the number providers advertise and it is the one people fixate on. It governs how fast pages load, how many streams you can run at once and how quickly large files arrive. Beyond a certain point it stops mattering: a single 4K stream needs a fraction of what most households buy.
Upload speed
How quickly data leaves your home. This is the number that determines whether video calls are stable, whether cloud backups finish, and whether sending large files is painless or infuriating. On cable and DSL it is a small fraction of the download figure; on fiber it matches. If you work from home, this is the number to watch, as our guide to internet for remote work explains.
Ping, or latency
The round-trip time for a small packet, in milliseconds. Low latency makes everything feel responsive. Under 30 milliseconds is excellent, under 60 is fine, over 100 is noticeable in interactive applications, and geostationary satellite sits around 500 or more because of the distance to orbit.
Jitter
The variation in latency between packets. This is the number nobody advertises and the one that most often causes problems. A steady 45 milliseconds feels far better than an average of 30 that swings between 10 and 90. Jitter is what makes video calls sound robotic and games feel like they are teleporting. Our gaming guide goes into why.
Packet loss
The proportion of data that never arrives and has to be resent. Anything above a fraction of a percent, sustained, will cause visible problems in real-time applications, even if download speed looks fine.
Reading your results
If wired speeds match your plan but Wi-Fi is slow
The connection is fine and the problem is your wireless network. The usual causes are router placement, an old router, thick walls, or interference from neighboring networks. Moving the router to a central, elevated, unobstructed position helps more than people expect. A mesh system is the reliable fix for a larger home. Upgrading your internet plan will not help at all.
If wired speeds are well below your plan at all times
This is a provider or equipment issue worth raising. Before calling, check that your modem or gateway supports the tier you are paying for, since an older modem can cap a faster plan. Then report it with your test results, times and dates.
If speeds are fine off-peak and poor in the evening
This is congestion. On cable it usually means a busy node in your neighborhood; on fixed wireless it means a busy tower. Reporting it is worthwhile because providers do track and address node congestion, but the honest answer is that it may not be quickly fixable, and it is a legitimate reason to consider a different technology. See fiber vs cable vs DSL.
If latency is high everywhere
Check what technology you are on. High latency is intrinsic to geostationary satellite and elevated on DSL and fixed wireless. On cable or fiber it usually indicates a fault or a badly configured router.
A speed test measures the path between your device and one test server at one moment. Consistently poor results across several services, several times of day and over a wired connection is evidence. A single bad result is not.
Why your speed test never hits the advertised number
Advertised speeds are maximums, and a small shortfall is normal. Some of it is protocol overhead, some is the test server, some is your own device, and some is that the last few percent of a connection is genuinely hard to measure. A result within roughly ten percent of your plan over a wired connection is healthy. A result at half your plan, consistently, is not. If the wired result is fine and only the far rooms are slow, the answer is coverage hardware, and our comparison of mesh Wi-Fi versus a Wi-Fi extender covers the choice.
It is also worth remembering that speed tests measure the connection, not the service you are complaining about. A slow streaming service on a fast connection is usually a problem at the other end.
Testing before you switch
If you are trialing a new provider, particularly fixed wireless, test it properly before canceling the old one. Run tests at your actual usage times for at least a week, pay attention to jitter and evening performance rather than the best single result, and check any work applications you depend on. Because most fixed wireless plans have no contract, this is a genuinely low-risk experiment — see our cable versus 5G comparison for how to structure it.
And before you conclude you need a faster plan, check whether you need one at all. Our guide to picking the right speed for your household works through what different activities genuinely consume, and the answer is usually less than people assume.