Every experienced technician owns one unglamorous superpower: they change one thing, test, and write it down. It feels slow for ten minutes and saves three days - the video case took the scenic route precisely because the first swaps were done before the setting was checked.
In the video case, the team changed the wire first. The drops continued. That was not wasted work - it was a test with a result: the patch lead moved down the suspect list, and attention correctly moved to the IO jack, the patching, and finally the adapter setting. Swap testing only fails when it is done carelessly - several changes at once, no notes, no re-test. Done properly, it can isolate any wired fault to one segment with parts you already own.
Example only: if the fault follows the wall point and not the computer, stop reinstalling drivers - the run, the jack or the switch port owns the problem.
After every single swap, re-run the same test: negotiated link speed plus 200 pings to the gateway. Same test, same numbers, comparable results.
One change per round. The moment you swap the cable and the port together, a cure teaches you nothing. Retest the same way every time - same target, same packet count. Write it down - intermittent faults outlast memory; a five-line log per round is the difference between a method and a wander. Restore as you go - put the original lead back before testing the port, or your baseline drifts and the matrix lies.
The video case carries the warning honestly: the team did correct physical work - wire, IO jack, re-patching - while the actual culprit sat in the adapter properties as a forced Speed and Duplex value. Physical swaps cannot find a settings fault. So the discipline is: read the link speed and the adapter settings before the first swap, and re-read them after any driver install or Windows imaging, because images and old templates love to carry forced settings from another era.
If every PC on one switch drops together, the shared box or its uplink is the suspect - swap testing per-desk is wasted motion. If drops hit the whole building at the same times of day, think power, heat and scheduled loads, not cables. The matrix isolates single-link faults; use the ping pattern to decide whether you even have a single-link fault before you start.
See the video - the wire swap that did not fix the Tally PC, and why that result pointed the team in the right direction anyway.
Watch Full VideoNo - it was a test with a clear result: that cable was probably innocent. The mistake would be changing three more things without re-testing. Move one step along the chain - jack, port, then settings - re-running the same ping test after each step.
Cross-test. Put a known-good laptop on the suspect wall point and the suspect PC on a known-good point. If loss follows the point, infrastructure is guilty; if it follows the PC, the adapter, its driver or its settings are. Two tests settle what weeks of guessing cannot.
The one change made, the negotiated link speed after it, and the ping result - sent, lost, and the pattern. Five lines per round. With an intermittent fault, that log is the only thing that survives until the pattern becomes obvious.