Faster Wi-Fi starts with locating the bottleneck. The internet service, router, wireless link, mesh backhaul, client device and application all contribute to the result. Compare the same task under controlled conditions, change one thing, and repeat the comparison.
You need administration access only for configuration changes. Initial tests need a laptop or phone, a consistent test service, and ideally an Ethernet-capable computer. Keep employer-managed security settings in place and record when a VPN is active.
Establish three comparison points
Choose a quiet period and pause optional cloud backups, game downloads and streaming on your own network. Keep the same test device, test destination and power mode. Run three measurements at each comparison point, a short interval apart, recording every result instead of keeping only the fastest.
- Wired baseline: connect the computer to a LAN port on the router. Temporarily turn off that computer’s Wi-Fi so the path is unambiguous. Check that the Ethernet adapter and cable support the speed being investigated.
- Nearby Wi-Fi: unplug Ethernet, enable Wi-Fi, and test a few steps from the router in an open location.
- Problem location: repeat on the same Wi-Fi network at the desk or room where the issue occurs. Keep the laptop in its normal orientation.
These internet tests include the provider and test server. They compare paths; they do not measure the Wi-Fi radio alone. A router’s built-in speed test may measure only its connection to the internet, excluding the laptop’s Wi-Fi link. Label it separately.
Use the downloadable test log to preserve conditions, results and your next action. It includes units, a filled fictional example, and blank records you can copy into a text editor.
Interpret the pattern before adjusting settings
On a narrow screen, scroll the table sideways. Keyboard: focus the table and use the arrow keys.
| Observed pattern | What it suggests | Useful next check |
|---|---|---|
| Wired and nearby Wi-Fi are both slow | A shared bottleneck: service, router, test destination or background traffic | Try another destination and device; check provider status and Ethernet link rate |
| Wired is good; nearby Wi-Fi is poor | Client or local wireless issue | Try a second Wi-Fi device, then inspect band and driver/firmware status |
| Nearby Wi-Fi is good; office is poor | Coverage, interference or mesh backhaul issue | Move the device or node and repeat the same task |
| Only one device struggles | Device-specific settings, software or radio limitations | Compare it with another device at the same spot |
| Speed looks adequate but calls break up | Delay variation, loss, congestion or application/device issues | Inspect call statistics during the problem |
These patterns narrow the investigation. For example, poor wired results could come from a 100 Mbps adapter as well as the provider. A successful second device makes a client problem more likely, while its different antenna and radio still affect the comparison. Microsoft’s connection troubleshooting includes comparisons across devices and frequency bands, plus the Windows 11 Get Help troubleshooter.
Worked example: a location problem
All numbers here are invented measurements in megabits per second (Mbps), using one laptop and one test destination:
On a narrow screen, scroll the table sideways. Keyboard: focus the table and use the arrow keys.
| Connection | Three download results, Mbps | Median, Mbps | Three upload results, Mbps | Median, Mbps |
|---|---|---|---|---|
| Wired router LAN | 310, 298, 305 | 305 | 22, 21, 22 | 22 |
| Wi-Fi near router | 270, 260, 268 | 268 | 21, 21, 20 | 21 |
| Wi-Fi at office desk | 38, 42, 35 | 38 | 8, 9, 7 | 8 |
| Office after moving the mesh node | 145, 152, 148 | 148 | 19, 20, 19 | 19 |
For three readings, sort them and take the middle value. Moving the node increased the office download median from 38 to 148 Mbps, about 3.9 times the original reading. That is an outcome of this fictional test, not an expected improvement for another home.
The useful conclusion is that changing the local path helped. Repeat at a busy time and run the actual video call before accepting the new position. The mesh guide explains why a node’s connection to the router matters as much as its connection to the laptop.
Make changes in a useful order
Start with reversible physical changes: move the router into the open, improve a mesh node’s intermediate position, or use an existing Ethernet run. Next, install supported firmware and the computer manufacturer’s current network driver. Record the original settings before changing channels or other radio options.
If the equipment exposes band information, compare an available alternative at the problem spot. Let its automatic channel selection provide a baseline. Nearby access-point counts alone say little about how much airtime those networks actually consume. Wider channels can also occupy more shared spectrum; a maximum-width setting deserves a measured comparison.
Avoid changing transmit power, security mode, channel width and node placement together. Even if the result improves, you would lose the evidence needed to explain it or recover from a regression. Schedule disruptive router changes when others can tolerate a short interruption.
Check delays and decide when to escalate
Repeat a real call or other interactive task under normal household load. Download throughput is measured in Mbps; delay is measured in milliseconds (ms). Jitter describes variation in delay, and packet loss describes missing packets. A high download result can coexist with noticeable call interruptions. See wireless video conferencing for a task-based test.
If every suitable wired device performs poorly at multiple times, send the provider your dated results, service plan, gateway model and test conditions. If only one laptop fails, give its support team the OS, adapter and driver versions plus the comparison result. Share logs through the official support channel after checking them for addresses, network names and other private details.
Upgrade when the test identifies a constraint the new equipment actually addresses: missing coverage, a limited wired port, unsupported clients or overloaded capacity. Retest after installation using the same record.