What a proper diagnosis looks like, so you can tell one from a guess

The complaint gets verified, not accepted

The first move is not to open anything. It is to establish with an instrument what the appliance is actually doing, because a real share of these calls is not a failure at all: a door held ajar by a bottle in the rack, a load of warm groceries an hour earlier, a setpoint somebody changed, a unit that was moved and never leveled. On a Sub-Zero built-in the verification is quick and it cuts the machine in half, because each compartment has its own sealed system and no air passes between them. A freezer still holding hard ice has already proved that power and one entire refrigeration circuit are healthy, and the fault belongs to the other side. It does not clear the control, since one board runs both compartments and can lose an output to the refrigerator side while the freezer carries on perfectly. That single check eliminates more possibilities than anything that follows it, and skipping it is how a visit ends with a compressor quoted on a machine whose compressor is fine.

• A calibrated probe in each compartment, given time to settle instead of read the moment the door shuts
• Which compartment is failing and whether the other is still holding, the fastest elimination this equipment offers
• Whether the fault is present right now or intermittent, because those are diagnosed in opposite ways
• What changed in the days before it started: a move, a power event, remodeling work, an unusually warm load

The order of elimination is not arbitrary

Faults are worked through from the most common and cheapest toward the rarest and most expensive, and the reason is not thrift. It is that each stage changes what the next stage's readings mean. A condenser packed with dust, or corroded down to bare aluminum on a unit that has spent years near the water, makes a perfectly healthy sealed system look weak on every measurement available, so judging the refrigerant charge before that coil is clean produces a confident, expensive, wrong answer. The same logic runs down the whole list. Air movement is judged after heat rejection. Sensing is judged after air movement. The control is judged last, once everything it depends on has been proved. A technician who opens with the sealed system on the first visit is not being thorough; they are skipping the work that would have shown they did not need to be there.

• Heat rejection first: the condenser, the clearance behind the grille, the condenser fan and its current draw
• Air movement second: the evaporator fan, the defrost circuit, and any ice built up behind the rear panel
• Sensing third: thermistors read as resistance and compared against the table at the temperature they are sitting in
• Switching fourth: door switches, relays, heaters and the wiring between them
• The sealed system last, named only when everything above reads healthy and still does not explain the cabinet

What the measurements sound like

You do not have to take these readings yourself, but you should be able to recognize that somebody took them. A diagnosis carries numbers, and every number has a reference it was compared against. A thermistor is a resistance in ohms compared with the manufacturer's table at the temperature the sensor is actually sitting in, which is a value and the value it should have been, not the phrase that the sensor tested fine. A defrost heater is checked for continuity and then for the current it genuinely draws, because a heater can read continuous and still be far too weak to clear a coil. A fan motor is a current reading and a check for bearing drag, not a glance at whether it spins. A door gasket is a strip of paper closed in the door at half a dozen points around the frame, and the places it pulls out freely are the fault. A drain is proved by pouring water through it and watching it arrive in the pan, not with a wire and an opinion.

• Ohms across a thermistor, against the table, at a known temperature
• Continuity and then current draw on a defrost heater, where the second test is the one that finds a weak element
• Clamp-meter current on the condenser and evaporator fan motors
• A paper pull test at several points around the door frame rather than one
• Water poured through the drain and confirmed arriving in the pan
• On board-managed units, the service-mode values read out and checked against an independent probe

The patterns that mean nobody knows yet

Some things are simply tells. Three parts replaced in one visit is not thoroughness, it is a parts cannon fired at a fault nobody has identified, and you pay for all three. A fault handed over as a system rather than as a part is another: a cooling problem is a category, the refrigerator evaporator fan motor has failed is a diagnosis, and only one of those can be checked once the visit is over. So is a reading with no reference attached to it, because the sensor tested fine is an impression, while a resistance in ohms set beside the figure the table wanted at that temperature is a measurement. And the most useful test an owner can apply costs nothing: ask whether the proposed repair actually explains the symptom you called about. If ice was cleared from behind the rear panel and nothing was found to explain why the defrost cycle stopped finishing, the ice is coming back, and everyone standing in the kitchen already knows it.

• Several parts replaced on one visit, none of them individually justified by a reading
• A fault named as a system rather than as a part and a failure mode
• A reading offered without the figure it was compared against
• Ice cleared with no explanation of why the defrost stopped finishing
• A repair that does not account for the symptom you actually reported

What you should be left holding

At the end of a real diagnosis you have the model and serial number written down, the fault named as a part and a failure mode rather than as a system, the readings that support it, and one price for the repair agreed before anything is fitted. Named as a part means the refrigerator evaporator fan motor has failed, not that there is a cooling problem. You should also be told what happens if you wait, because that varies enormously: a weak evaporator fan costs you groceries and time, while a compressor short cycling against a restricted condenser is spending the compressor. Honest answers include unwelcome ones. An intermittent fault that is not present on the day genuinely cannot be closed out with a part, and being told it needs to be watched and measured is a better answer than a confident guess. Occasionally the honest answer is that a particular machine is not worth the repair, and a technician who says so has done you a favor rather than lost a sale.

Asking what was measured, before the part goes in

None of this requires suspicion, and asking is completely normal. What did you measure, and what should it have read? That separates a reading from an impression, and anyone who took one answers it in a sentence. What else can produce this symptom, and how were those ruled out? Every fault has a differential, and somebody who worked through it can list theirs without pausing. And if this part goes in and the fault is still there, what comes next? That one is far better heard before the part is fitted than after, because it is what shows whether the diagnosis is a conclusion or a first attempt. All of it is the same reasoning that gets done internally on the way to an answer, which is why it is easy to answer out loud when the answer was actually reached.