How to Run an Accurate Internet Speed Test (and Read the Results Correctly)

August 21, 2026 · Fiber internet for real communities

Public domain, via Wikimedia Commons

Speed tests seem simple. Click a button, get a number, compare it to your bill. But most people run the test wrong and then misread what it tells them. Part of the confusion comes from the test itself, which measures a lot more than just your internet plan. The rest comes from marketing that treats "speed" as one clean number when it is really a handful of different measurements that all matter for different reasons.

"My speed test result is exactly what my internet plan delivers"

This is the biggest misunderstanding. A speed test measures the connection between your device and a test server at that exact moment, over whatever path your traffic takes to get there. That path includes your WiFi signal, your router, your device's own hardware, and how far away the test server is. None of that is your "internet plan." Your plan is the speed your provider delivers to the equipment at your home or business. What the test shows you is what survived the trip from there to your laptop or phone. The kernel of truth is that a well run test on a good connection should land close to your plan speed. It just is not automatic.

"I should test on WiFi because that's how I actually use the internet"

This makes intuitive sense but it backfires. If you want to know what your internet service is actually delivering, test with a device plugged directly into your router or ONT with an ethernet cable. WiFi adds variables: distance from the router, walls, other devices on the network, interference from neighbors' networks, and the age of your router's WiFi radio. A wired test isolates your actual internet speed. Once you've confirmed that number is right, then it makes sense to test on WiFi separately to see how much your wireless setup is costing you. Testing only on WiFi conflates two different problems and makes it hard to tell whether to call your provider or fix your router placement.

"One test at one time is enough to know my real speed"

A single test tells you what happened in that ten-second window, on that server, at that moment. Networks are not static. Speeds can shift based on time of day, especially in the evening when everyone in your area is streaming or gaming at once. The test server you land on matters too. If it is far away or congested, your result will look worse than your connection actually is. Reputable test tools let you pick a server or will auto select the closest one, but it is still worth running the test a few times, at different times of day, and comparing results. If you get 900 Mbps at 10am and 200 Mbps at 8pm on a gig plan, that pattern tells you something a single test cannot.

What actually matters instead

Focus on three things: how you test, when you test, and what you are actually checking for.

• Test wired first. Plug a laptop or desktop directly into your router or ONT to get your baseline internet speed, separate from WiFi issues.
• Run it more than once. Test at different times of day, especially during evening peak hours, and note any big drops.
• Look at upload and latency, not just download. Download speed gets all the attention, but if you work from home, run video calls, upload large files, or run a business with cloud tools, upload speed and latency (ping) matter just as much, sometimes more. A connection with high download but poor upload will feel fine for watching video and frustrating for video calls.
• Compare against your plan honestly. Getting 900 Mbps on a 1,000 Mbps plan is normal and expected, not a problem. Overhead from network protocols means you rarely see the exact advertised number, and that's fine. A meaningful gap, like getting 300 Mbps on a 1,000 Mbps plan consistently, wired, at different times of day, is worth a call to your provider.
• Know the difference between a WiFi problem and a connection problem. If wired speeds are good but WiFi speeds are poor, the fix is usually router placement, a mesh system, or an older router that needs replacing, not a call to your ISP.

Community and municipal leaders evaluating broadband in their area should apply the same logic at scale: wired tests, multiple times of day, across multiple locations, before drawing conclusions about whether a network is underperforming. A handful of WiFi tests run at noon will not tell you much about whether a neighborhood's fiber build is actually delivering.

At GigSpeed, we tell customers the same thing we'd tell a neighbor: test wired, test more than once, and expect numbers close to your plan speed, not identical to it. If you're doing that and still seeing a real, repeated gap, that's a conversation worth having with your provider.