← All guides

You pay for "up to 500 Mbps," run a speed test, and see 340. Run it again five minutes later and it says 280. Neither number is wrong, and neither one is really a measurement of your internet plan on its own — here's what's actually happening between your ISP's marketing page and the number a speed test shows you.

What "up to" actually means

ISP-advertised speeds describe the maximum your connection is provisioned for under ideal conditions — a ceiling, not a guarantee. Real conditions are almost never ideal: other households on the same neighborhood node competing for capacity, the specific routing path between you and whatever server is doing the measuring, and normal protocol overhead all sit between that advertised ceiling and whatever number shows up on your screen. A result meaningfully below the advertised figure isn't automatically a sign something's broken.

Where the gap actually comes from

A few concrete things quietly eat into a result before it ever reaches a speed test:

• Wi-Fi, not the connection itself. Wi-Fi has its own, separate throughput ceiling that's often well below what your wired connection could deliver — a router on the older 2.4GHz band, sitting two rooms away with a wall in between, can bottleneck a fast connection down to a fraction of what it's actually capable of.
• Every other device on the network. A phone backing up photos, a smart TV streaming in another room, a laptop mid-update — all of it shares the same pipe at the same time, whether or not anyone's actively watching it happen.
• Distance and routing to the test server. Data still has to physically travel, and more hops along the way means more chances for small delays to add up — which is part of why the same test can read differently depending on which server it happens to reach.
• TCP overhead and connection ramp-up. A fresh connection doesn't hit full speed instantly — it ramps up over the first moments of a transfer, so a very short test can undercount a connection that would otherwise sustain a higher number over a longer run.

Why two different speed test tools give two different numbers

Beyond your own network, the test itself is a variable. Different tools use different test servers in different physical locations, run for different lengths of time, and open a different number of simultaneous connections while measuring — more parallel connections generally push a number higher, since it's less exposed to any single connection's own overhead. None of these choices is more "correct" than another; they're just different methodologies, which is exactly why your ISP's own app, this tool, and a third one can all disagree on the same connection within the same minute.

What's actually worth paying attention to

A single number from a single test says less than a pattern across several. Testing at different times of day shows whether your connection holds up during your neighborhood's busiest hours, testing with other devices idle isolates your raw connection from whatever else is sharing it, and testing over a wired connection when possible rules Wi-Fi out as a variable entirely. A result that's consistently and substantially below what you're paying for, across several tests like that, is a real, useful signal worth raising with your ISP — a single lower-than-expected reading generally isn't.