No Fault Found: What Now Decision Tree

Why this matters

Sooner or later you finish testing and everything passes, yet the customer swears it failed. This is the moment that defines whether you are a parts-swapper or a diagnostician. Panic-replace something and you look like you are guessing, because you are. Shrug and leave and the fault returns with your name on it. "No fault found" is not a dead end. It is a specific situation with a specific set of next moves, and handling it well is one of the most professional things a tech does. This tree is those moves.

Start here: confirm there is no live hazard

Before you treat a call as no-fault-found, make sure you are not walking away from a hidden hazard. If the complaint involved any sign of gas, an electrical trip, overheating, or water near live parts, do not close it out as "nothing wrong" - de-energize or shut off and verify safe first. A safety symptom that "stopped on its own" is a reason to lock the equipment out, not to clear it. Only a complaint with no safety dimension belongs on the diagnostic path below.

First: did you actually reproduce the complaint

The single most common cause of no-fault-found is testing the wrong thing. Before you conclude nothing is wrong:

• Recreate the customer's exact conditions, not yours. They get the fault on the third cycle, in the morning, under full load, with a door shut. Did you test that, or did you test at idle for ten minutes?
• Ask them to show you if they can. Watching them operate it surfaces a use pattern you would never test for, and sometimes the fault is the use pattern.
• Match the variable. Hot, cold, loaded, after an hour, only when another load runs - if you did not test under the condition they describe, you have not tested the fault. You have tested a healthy machine doing nothing hard.

Most "no fault found" is "fault not reproduced." Fix that before you go further.

Second: was the fault real but already gone

Some faults clear themselves and leave evidence behind. Look before you declare nothing happened:

• Check for fingerprints. A tripped-and-reset protector, a logged error, a heat-discolored connection, a fresh water stain, a blown-then-replaced state. The fault was real; it just is not active right now.
• Read the history. A part that overheated and cooled, a connection that arced and reseated, a control that faulted and cleared. These are intermittents, not phantoms. They belong on the monitor path.
• Believe the customer's timeline. If they describe a clear, repeatable symptom, the fault is almost certainly real and you simply have not caught it. Treat it as an intermittent you must instrument, not a non-event.

Third: rule out the boring causes

Before anything exotic, clear the cheap explanations that masquerade as a fault:

• Customer operation - a setting changed, a mode selected, a control bumped, a supply turned off. Not a defect, but a real complaint with a real fix.
• Supply or environment - marginal voltage, low pressure, a tripped upstream device, a shared circuit overloaded only at peak. The unit is innocent; its supply is the problem.
• Normal behavior misread - a sound, a cycle, a temperature swing, or a runtime that is normal for the equipment but new or alarming to the customer. The fix is teaching, not parts.

A surprising share of no-fault-found calls land here, and the honest answer is "the equipment is fine, here is what you saw and why."

The branches

Do not panic-replace

When tests pass, the worst move is grabbing the most expensive likely part and throwing it at the problem to look decisive. That is how you charge for a non-fix and earn the callback anyway. If you genuinely cannot find it and the consequence is high, replace at most one low-cost most-likely item, say openly that it is a suspicion not a diagnosis, and leave instrumentation to catch the real fault. Resist the urge to do more.

Close it out professionally

End a no-fault-found call by leaving the customer in a better place than a shrug:

• State plainly what you tested and what passed.
• Explain what you think happened (not reproduced, intermittent, supply, or normal behavior).
• Say what triggers a return: it acts up again, or your recorder catches it.
• Write it all down so the next visit starts with this knowledge.

The judgment to bank: "no fault found" almost always means "not reproduced," "intermittent," or "boring cause." Work those three before you ever touch a part, and never let a clean test become a blind replacement.

References

• Trade-standard practice for fault reproduction and intermittent diagnosis
• Manufacturer documentation on normal operating behavior and error history
• OSHA general guidance on energized work and lockout/tagout for hazard branches (29 CFR 1910)
• See related: Everything Tests Good But It Fails; When to Tell the Customer You Can't Find It