A Wi-Fi test is a photograph of one moment, one chair, one laptop. Treat it that way and it is genuinely useful. Treat it as a specification and it will lie to you with a straight face.
Running iperf3 over Wi-Fi is easy - and easy to misread. The number you get is real, but it is the speed of the whole path at that moment: the laptop's radio, the air between it and the access point, the access point itself, and the wired network behind it. This guide shows how to run the test so the wireless part is what you are actually looking at, and states plainly what Wi-Fi results can and cannot prove.
If both PCs are on Wi-Fi, your test crosses the air twice, and the result is some blend of two radios and one shared access point. You will learn nothing clean from it. Always put the iperf3 server on a wired PC - ideally the actual server or a PC on the same switch as the access point - and test from the laptop on Wi-Fi.
Now the path contains exactly one wireless hop, and the number describes that hop plus a wired section you can verify separately in seconds. One variable at a time, or the test is theatre.
Wi-Fi speed is negotiated moment by moment based on signal quality. Walls, distance, a metal cabinet, a neighbour's network on the same channel, even the laptop's own orientation all change the rate. So a result is honest only with its context attached: which room, how far from the access point, which band.
Test on the 5 GHz band near the access point for the best case, then from the seat where work actually happens for the real case. Record both. The gap between them is your coverage problem, measured instead of guessed. Also run the test twice - Wi-Fi results wander, and quoting one lucky run is how bad purchases get approved.
An iperf3 result cannot prove what your Wi-Fi is rated for. A laptop linked at 866 Mbps will never show 866 in iperf3 - protocol overhead, one radio sending or receiving at a time, and real-world airtime sharing take a large cut. Roughly half to two-thirds of the link rate in a clean test is respectable, not broken.
It also cannot separate a weak laptop radio from a weak access point by itself. For that, test a second laptop from the same chair. If both are slow there and fast elsewhere, blame the spot or the access point. If only one is slow everywhere, blame the laptop.
Before condemning the wireless, test the wired side once: iperf3 between the wired server and another wired PC on the same switch. If that is not near 940 Mbps, your problem is not in the air at all.
And remember the direction trick from our reverse-mode guide: on Wi-Fi, -R matters even more, because a laptop usually receives far better than it sends. Test both directions from the problem chair before deciding anything needs replacing.
The complete iperf3 setup and live testing, step by step, in the video.
Watch Full VideoProbably not. The link rate is the radio's raw signalling speed before overhead. Real data throughput on Wi-Fi is typically half to two-thirds of that under good conditions, and less with distance or interference. Compare your result with a test near the access point before calling anything faulty.
Wired, always. The server side should sit on a cable - the real server or any wired PC - so the only wireless section in the path is the laptop you are testing. Two Wi-Fi ends mix two unknown radios into one unreadable number.
Because the air changed: a neighbour's traffic burst, a microwave, someone walking through the signal path, the access point shifting rates. Run each position two or three times and trust the typical result, not the best one.