Speed test says 300 Mbps. File copy crawls at 2 MB per second. Both numbers can be true at the same time, because the two tests never measured the same road.
This is the most common wrong turn in network troubleshooting. A user complains that the shared drive is slow, somebody opens a speed test website, the result looks wonderful, and the complaint gets closed as user error. It was not user error. The speed test measured the path from that PC out to the internet. The shared files live on a server inside the building. Those are two completely different roads, and only one of them is slow.
When you run a speed test in the browser, your PC connects to a server on the internet - often in a nearby city - and measures that round trip. The result includes your broadband line, your router's internet port, and your provider's network. It says nothing about the cable between your desk and the office switch, the switch itself, or the server in the corner.
Worse, many speed test sites quietly pick the friendliest nearby server and use several parallel connections, which produces the best-looking number the line can offer. That is fine for checking your internet plan. It is useless for a local file problem.
iperf3 runs between two machines you choose. Start it as a server on the file server with iperf3 -s, run it as a client on the user's PC with iperf3 -c 192.168.1.10, and the number you get is the speed of the exact path the shared files travel - same cables, same switch, same network cards.
If that path is broken, iperf3 shows it plainly. A gigabit office path returning 188 Mbps, or 94 Mbps, is a fact you can act on. No opinion, no feelings, no blame on the user.
The mistake in the video's office story is one engineers make everywhere: testing the healthy road and declaring the whole network healthy. The user limited to about 2 Mbps on local downloads was never going to be helped by an internet test, because their files never leave the building.
The honest rule: match the test to the journey. Files on a local server - test locally with iperf3. Website slow for everyone - then test the internet. Video call with the branch office - test the path to the branch. Each complaint has its own road.
Fairness matters: iperf3 is not always the answer. If the whole office says websites are slow but local files fly, the local network is probably fine and the internet line or provider is the suspect - an internet speed test, run at different times of day and kept as dated records, is exactly right there.
And if iperf3 between two office PCs is fast but one user's PC is still slow at everything, the problem may be that PC itself - its disk, its cable, its network card. The tests narrow the search; they do not replace thinking.
The complete iperf3 setup and live testing, step by step, in the video.
Watch Full VideoMost likely the local path: a cable or switch port running at 100 Mbps instead of gigabit, a Wi-Fi connection doing LAN work, or a busy or failing network card. Run iperf3 between the PC and the server - if it also shows a low number, the local path is guilty and the internet test was simply measuring something else.
No. iperf3 measures between two machines you control. It cannot tell you what your provider delivers to your building. Use each tool for its own road: iperf3 for local and branch paths, an internet test for the broadband line.
Start with iperf3 between that PC and the server, wired if possible. Then compare with a test from a known-good PC on the same switch. The gap between the two results tells you whether to suspect the user's cable and port, or the shared path and server.