Internet Speed Test IO

CONNECTION LAB · ENGLISH EDITION

Is slow internet caused by Wi-Fi, your device or your provider?

Separate the local wireless link from the internet connection with one controlled change at a time. This procedure produces evidence, not an automatic verdict against your provider.

InternetSpeedTest.io · Updated September 9, 2026 · Practical guidance, not a line certification

1. Establish a baseline you can repeat

Choose one device and one test service. Pause optional downloads, cloud backups and streams on your own devices. Keep the browser tab visible. Record whether a VPN is active; only disconnect it if it is optional and you are permitted to do so. Do not disable security software or workplace protections for this test.

Run three completed tests in your usual location, waiting for each to finish. Write down download, upload, ping, jitter and request loss separately. Keep failed attempts in your notes as failures, not zero-speed results. Use the median of the completed measurements and record the range.

Tests transfer real data. Check your allowance before repeating them on a metered connection. Stop if testing would interrupt an important call or use data you cannot spare.

2. Change location, not everything else

Move the same device near the router and repeat. Keep the same Wi-Fi network and test service. A large, repeatable improvement makes the wireless path a useful next place to investigate: walls, placement, interference or the link to an extender. It does not identify which of those is responsible.

If your router exposes separate bands, compare them while close to the router and label the band for every run. A higher-frequency band can offer more capacity nearby but is not automatically better through walls. Do not compare a busy evening run on one band with a quiet morning run on another and call that a band comparison.

3. Compare a wired connection, when available

Connect the same computer by Ethernet to your own router, and confirm it is using Ethernet rather than Wi-Fi. Repeat the baseline. If a different device is required, write that down: the comparison now includes a device change, so the result is less conclusive.

Check the negotiated link speed in the device's network settings. A 100 Mbps link can constrain a faster plan. A worn cable, a port limit or an adapter can be relevant, but a result near 100 Mbps alone does not prove any particular component is broken. Do not connect directly to an optical terminal or change provisioning settings without your provider's instructions.

Read the pattern

Observations and the next useful check — not definitive diagnoses
Repeatable observationNext checkWhat it cannot prove
Wired is stable; far-room Wi-Fi is poorRouter placement, band, extender pathThat buying a new plan will help
Only one device is slow near the routerThat device's radio, adapter and background trafficThat the provider is throttling it
Several wired devices are slow at busy timesRepeat at a quiet time; log household trafficWhether congestion is at home or upstream
Only one website or game is slowIts service status and route, using its own diagnosticsThat the whole internet connection is slow

Worked example: avoid the wrong upgrade

Illustrative values, not measured customer data: three bedroom runs give 22, 26 and 24 Mbps; near-router runs give 91, 95 and 93 Mbps; wired runs give 94, 96 and 95 Mbps. The medians are 24, 93 and 95 Mbps. This pattern makes the bedroom wireless path worth investigating before paying for a faster plan.

It still does not establish the line's maximum capacity. The wired adapter or service under test might impose another limit. After moving the router, repeat the bedroom test under comparable conditions rather than relying on the attractive near-router result.

Prepare a useful support request

Include the affected activity, dates and local time zone, device, wired or wireless connection, all completed results and failed attempts, and the checks you tried. Ask the provider to investigate repeated wired degradation or disconnections. Do not send Wi-Fi passwords, router administrator credentials or unredacted account information in a public forum.

A concise report might say: “Video calls break up between 20:00 and 21:00. I paused backups and repeated three tests over Ethernet. Here are the results and times. The same device was stable earlier.” This describes evidence without assigning an unproven cause.

Calculate the median and prepare your own notes →