How to Test Your Internet Speed Correctly
A speed test is a snapshot: it measures your connection at one moment, from one device, to one server. With a few simple steps you can make that snapshot as fair as possible, and learn far more from it.
Before you start
- Pause other internet use. Stop downloads, game and system updates, cloud backups and video streams on all devices in the home. Anything running in the background takes a share of the connection.
- Close other tabs and apps on the device you are testing with. Browser tabs can keep loading data quietly.
- Turn off your VPN, unless you specifically want to measure performance through the VPN. VPNs add distance and processing, and usually lower both speed and latency results.
- Keep the test page in the foreground. Browsers slow down background tabs to save power, which can lower results. SpeedCheckWeb warns you if the page was hidden during a test.
- Plug in a laptop. Some devices reduce network performance on battery-saving modes.
Cable first, then Wi-Fi
If possible, run two tests:
- Wired (Ethernet) test: connect a computer directly to your router with a cable. This shows what your internet connection delivers, without Wi-Fi limits.
- Wi-Fi test: test from the places where you actually use the internet, such as the sofa, the home office or the bedroom.
If the wired result is good but Wi-Fi is much slower, the problem is inside your home, not with your provider. Our Wi-Fi guide can help.
Test more than once
Results naturally vary from test to test. A good approach:
- Run three tests in a row and look at the typical result, not just the best or worst.
- Test at different times of day: morning, afternoon and the evening peak (about 7–11 pm in many areas). If speeds drop sharply only in the evening, local congestion is a likely cause.
- Keep a record. SpeedCheckWeb saves your results in your browser’s test history, and you can export them as a CSV file to share with your provider.
Read the whole result, not just the big number
| Result | Look for |
|---|---|
| Download | Close to your plan’s speed on a cable connection. |
| Upload | Close to your plan’s upload speed. This is often much lower than download. |
| Idle latency | Low and steady. Under about 50 ms to a nearby server is good. |
| Jitter | Low: under about 15 ms is good for calls and games. |
| Latency under load | Should not rise much above idle latency. A big rise means bufferbloat. |
| Packet loss | Ideally 0%. Shown only when it could actually be measured. |
Know what a speed test cannot tell you
- Every test measures a path to one server. SpeedCheckWeb tests to the nearest Cloudflare data centre. Speed to a game server, a video service or a website on another continent can be different.
- Your device can be the bottleneck. Older phones, budget laptops and USB network adapters may not reach gigabit speeds. A test in a web browser on a very fast line can also be limited by the browser itself.
- Plans are “up to” speeds. Providers advertise the maximum speed of a plan. Some overhead is always used by the network protocols, so even a perfect line measures a little below its headline speed.
- A test uses data. If you are on a metered mobile plan, turn on SpeedCheckWeb’s Data saver option.
Quick checklist
- Other devices paused.
- VPN off (unless you are testing it).
- Test page in the foreground.
- One wired test, plus Wi-Fi tests where you use the internet.
- Three tests, at more than one time of day.
- Results saved.
Start a speed test now or try the ping and bufferbloat test.
Sources
Figures from third parties can change; we review guides regularly. See our editorial policy.