How to Test Your Broadband Speed Accurately: A UK Checklist

A single speed test on a phone, on the sofa, while someone streams in the next room, tells you very little. It might be useful as a quick check, but it won't tell you whether your provider is delivering what you pay for, and it won't stand up if you need to complain.

This checklist takes about 15 minutes per session, and gives you results you can rely on and that your provider will take seriously. It also explains what each number means and how to compare it with the Minimum Guaranteed Speed in your contract.

Why Speed Test Results Vary So Much

The spread in real-world results is enormous. Across our tests, a tenth came in below 27 Mbps and a tenth above 850 Mbps, with the median at 151 Mbps. Even within one provider the range is wide: BT tests ran from 16 Mbps at the 10th percentile to 724 Mbps at the 90th (1,828 tests), because BT customers range from long copper lines to gigabit full fibre.

Your own results will vary too, for reasons that have nothing to do with your line: Wi-Fi, a busy household, a VPN, an old laptop, or the time of day. The steps below remove those variables one by one, so the number you end up with reflects the connection itself. Our test, like most browser tests, measures the whole path from your device to a nearby server, so anything in that path counts.

Step 1: Restart Your Router

Turn the router off at the wall, wait 30 seconds, and turn it back on. Give it five minutes to reconnect and settle. Routers that have been running for weeks can slow down as memory fills, and a restart rules that out. If you have full fibre, leave the separate fibre box (the ONT) alone unless your provider tells you to restart it. Don't restart repeatedly during testing, though: on some part-fibre lines, frequent reconnections can make the network's line management lower your speed for stability.

Step 2: Plug In With an Ethernet Cable

This is the single biggest improvement you can make to accuracy. Connect a laptop or desktop directly to one of the router's Ethernet ports, not via a switch, a mesh satellite or a powerline adapter.

In our data, tests our system classed as likely wired had a median download of 651 Mbps and ping of 10 ms; likely Wi-Fi tests had 142 Mbps and 42 ms. Not all of that gap is Wi-Fi (wired testers tend to have faster packages), but much of it is. Our Wi-Fi vs Ethernet guide breaks the comparison down.

Check the cable and port too. A gigabit Ethernet port tops out at roughly 940 Mbps of real throughput, older 100 Mbps adapters cap you at about 94 Mbps, and a damaged cable can silently drop to 100 Mbps. If you're on a package faster than 1 Gbps, you'll need a 2.5 Gbps port to measure it.

Step 3: Stop Everything Else Using the Connection

• Pause streaming on TVs and tablets, and check nobody is gaming or on a video call.
• Pause game downloads and console updates, which often run in standby.
• Pause cloud sync and backup: OneDrive, iCloud, Google Photos, Dropbox, Time Machine and similar.
• Close other browser tabs and apps on the test device.
• If you can't pause everything, use the router app to check which devices are busy, and note it alongside your result.

Step 4: Turn Off Your VPN

A VPN routes your traffic through another server and encrypts it, which costs speed and adds latency. Work laptops often run a corporate VPN or security client that does the same. Disconnect it, or use a personal device. Our VPN speed loss guide shows how big the hit can be. (We filter known VPN and data-centre results out of our published statistics for this reason.)

Step 5: Run Three Tests and Take the Middle Result

Run the test three times in a row and use the middle download result, not the best one. One test can be thrown by a brief burst of traffic; the middle of three is far steadier. That's the same reason we publish medians rather than averages for every provider and town on the site.

Step 6: Repeat at Different Times of Day

Do a session in the morning and another in the evening peak, between 8pm and 10pm, ideally on at least three different days. If your speed is fine in the morning and drops sharply every evening, that points to congestion on your provider's network rather than a fault on your line.

In our data, Openreach-based providers were about 13% slower at peak than at other times, and median ping across all tests rose from 37 ms overnight to 45 ms in the evening. Our broadband speed by time of day analysis has the hour-by-hour figures.

Step 7: Keep a Record

For each session, write down or screenshot:

• Date and time
• Wired or Wi-Fi, and which device
• Download, upload, ping and jitter from each of the three tests
• Anything unusual, such as other devices active or a recent router restart

A week of this is exactly the evidence your provider will ask for. It also tells you whether a problem is constant, evening-only or random, which changes the fix.

What Each Result Means

A few rules of thumb when reading your own figures:

• Download near 80 Mbps and upload near 20 Mbps is the ceiling of part fibre (FTTC). If that's you, you're probably getting the most your line can do. See FTTP vs FTTC.
• Ping above 50 ms on a fixed line is on the high side. About 27% of all tests in our data were above it, and almost all of those (97%) were classed as likely Wi-Fi.
• Loaded latency far above idle ping means another device filling the line will cause lag. Our ping guide explains why.
• Mbps is not MB/s. Divide by eight to get the megabytes per second your downloads show. See speed vs bandwidth.

Checking Results Against Your Minimum Guaranteed Speed

When you sign up, most UK providers give you three speed figures for your specific line: an estimated range, a typical peak-time speed, and a Minimum Guaranteed Speed (MGS). You'll find them in your contract summary, welcome email or online account. This comes from Ofcom's voluntary Broadband Speeds Code of Practice, in its current form since 2019, which most large providers have signed. Check Ofcom's website for the current list of signatories.

The MGS is a download speed and is usually well below the headline figure on the advert. It is meant to be the floor, not the norm. To compare fairly:

1. Use wired results only. Providers measure at the router, and Wi-Fi shortfalls aren't covered.
2. Use the middle of three from each session, and look across several days.
3. Compare evening results with the peak-time figure you were given, and all results with the MGS.
4. Run your provider's own speed checker too. They will usually ask for it, and it removes any argument about the test server.

If your wired results are consistently above the MGS but below the advertised speed, you have no right to exit under the code, but you do have evidence for a conversation about a cheaper or more suitable package.

What to Do if You're Below It

1. Report it to your provider with your record of wired tests, dates and times. Ask them to run a line test.
2. Give them time to fix it. Under the code, signed-up providers have 30 days to resolve speeds that fall below the MGS.
3. If they can't, you can leave. Under the code, if the problem isn't fixed in that time, you can exit your contract without early termination charges, and that has generally extended to landline and TV bought in the same bundle. Get the provider's confirmation in writing before you cancel.
4. Escalate if needed. If your complaint isn't resolved after eight weeks, or the provider sends a "deadlock" letter, you can take it to the provider's alternative dispute resolution scheme (Ombudsman Services: Communications or CISAS) for free.

These rules and processes can change, so check the current details on Ofcom's website and in your provider's complaints code. Our guide to complaining about broadband speed walks through the process, and why is my broadband slow covers the fixes worth trying first. You can also compare your results with other customers of your provider on the ISP rankings.

FAQ

Why do different speed test sites give different results?

They use different servers in different places, different numbers of simultaneous connections and different ways of summarising the measurements into one figure. Stick to one test for your records, and use your provider's checker as well when you complain.

Is a speed test on my phone accurate?

It's accurate for what your phone gets over Wi-Fi in that spot, which is useful to know. It isn't a fair measure of your broadband line, and providers won't accept it as evidence of a line fault.

How many times should I test my broadband?

Three tests per session, in at least two sessions a day (morning and evening), over three or more days. That's enough to tell a constant problem from an evening-only one.

Why is my speed lower than my package?

Package speeds are typically sold as an average or "up to" figure measured at the router, and a wired test usually lands somewhat below it because of network overheads. Wi-Fi, busy households and evening congestion take it lower still. Only consistently wired results below your Minimum Guaranteed Speed indicate a problem your provider must fix.