Watch an experienced technician work on a dead machine and it looks like intuition. It mostly isn't. It's a method — one so consistent you can learn its skeleton in an afternoon, then spend a career filling in the details. The method has three rules.
Rule 1: Establish what actually happens
"It's dead" describes at least four different faults. Before touching anything, observe precisely: Do fans spin? Any lights? Any beeps? Does anything appear on screen, even briefly? Each answer eliminates whole categories. Fans spinning but no display is a completely different investigation from total silence — the first suggests graphics, RAM, or display; the second points at power delivery. Vague symptom, vague diagnosis. Precise symptom, short list.
Rule 2: Simplify to the minimum, then rebuild
A computer that won't start with six components attached might start with three. Strip to the minimum bootable configuration — motherboard, CPU, one stick of RAM, power — and test. If it lives, add components back one at a time, testing after each, until the fault returns. Whatever you just added is your suspect. If even the minimum won't start, your suspect list is already down to three or four items.
The discipline that separates good troubleshooters from frustrated ones: change one variable at a time. Swap the RAM and reseat the graphics card and update the BIOS, and when the machine boots you've learned nothing — you can't tell which change fixed it, so you can't prevent it recurring or explain it in the ticket.
Rule 3: Swap with known-good, not with guesses
The fastest way to convict a component is to replace it with one you know works — and the fastest way to clear it is to try it in a machine you know works. This is why workshops keep a shelf of boring, verified spares: a PSU, a RAM stick, a display cable, a spare drive. Two swap-tests often close an investigation that hours of speculation couldn't.
The cheat sheet of prior probabilities
Method beats memorisation, but knowing what usually fails helps you order your tests. Roughly, from most to least likely: cables and connections (reseat everything first — it's free), storage drives (the moving-parts and wear-out champions), power supplies, RAM, fans and cooling, and only then, rarely, CPUs and motherboards. Statistically, "it's probably the cable" is the most useful sentence in hardware support — check the free things first.
None of this requires deep electronics knowledge. It requires observation, one-variable discipline, and a small shelf of known-good parts. The encyclopaedia comes later, on its own, one solved ticket at a time.