Good-Enough Diagnosis vs Perfect: Decision Tree
Why this matters
Chasing certainty on every call sounds like good craftsmanship, but it can be its own failure. Spend an hour proving the exact root cause of a worn-out part that obviously wore out, and you have burned the customer's money and your day for a confirmation that changed nothing. The opposite trap is just as real: declaring "probably the X" and swapping it on a hunch, when ten more minutes of testing would have caught the real cause. The skill is knowing when you have diagnosed enough to act safely and correctly, and when "good enough" is actually not. This tree calibrates that line.
Safety first: there is no good-enough on a hazard
One category never gets the good-enough treatment: anything that could be a hazard.
- If the fault involves gas, water near electricity, overpressure, overheat, stored energy, or a safety device, you confirm the condition is safe before you stop. "Probably fine" is not a diagnosis on a hazard.
- Lead with the safety action, verify the actual condition, and only then judge whether the rest of the diagnosis needs to go deeper.
You can be pragmatic about effort on a comfort complaint. You cannot be pragmatic about whether something might hurt someone. Hazards get certainty, full stop.
Start here: what does the diagnosis decide?
The depth you need depends on what rides on being right.
- It decides a low-cost, reversible repair (a cheap, common part that obviously failed): a good-enough diagnosis is fine. If you are wrong, the cost of the miss is small and you can recover.
- It decides an expensive, irreversible action (a major component, a teardown, a recommendation to replace the whole unit): the diagnosis must be solid. The cost of being wrong is large and hard to undo.
- It decides whether something is safe: see above. Always certainty.
Match your diagnostic effort to the stakes of the decision it drives. Over-diagnosing a trivial repair wastes money; under-diagnosing a major one wastes far more.
The good-enough test
You have diagnosed enough when all of these are true.
- The evidence points one clear way and nothing contradicts it. You are not ignoring a finding that does not fit.
- The failed part's condition matches the symptom. A part that obviously failed from normal wear, with a symptom that matches, does not need its root cause proven to a certainty.
- Nothing upstream is hunting the new part. You have ruled out a cause that would just kill the replacement. See the related cheap-fix tree.
- The repair is reversible or low cost if you turn out to be wrong.
When all four hold, proving more is effort without payoff. Do the repair.
When good-enough is not enough
Push for a deeper, confirmed diagnosis when any of these is true.
| Signal | Why you keep going |
|---|---|
| Findings contradict each other | Something does not fit; a wrong call is likely |
| The part is a victim, not a culprit | The cause will take out the replacement |
| The action is expensive or irreversible | The cost of being wrong is high |
| It is a callback on a prior fix | The first diagnosis was already wrong once |
| Anything safety-related | Certainty is mandatory |
A callback deserves special mention. If you are back on a fault you already "fixed," the good-enough diagnosis you did last time was not good enough. Do not repeat it faster; go deeper.
Beware the two failure modes
- Analysis paralysis: testing past the point of decision, proving things that do not change the repair, because stopping feels like cutting corners. On a clear, low-stakes fault, this is waste, not diligence. See the related analysis-paralysis tree.
- False confidence: stopping at the first plausible cause without checking whether the evidence actually supports it, because you are in a hurry. On a high-stakes or contradictory fault, this is how guessing masquerades as diagnosis.
Good techs drift toward one or the other. Know which way you lean and correct for it. The careful tech who over-tests should ask "does this next test change what I do?" The fast tech who under-tests should ask "would I bet the callback on this?"
Document the line you drew
Whichever way you call it, note in the job record what you confirmed and what you reasoned but did not prove. If you did a good-enough diagnosis on a clear wear failure, say so. If you stopped short because the stakes were low, that is defensible and on record. The note protects you: it shows the decision was a judgment, not a corner cut in ignorance.
The recap
- Hazards always get certainty. No good-enough on safety.
- Match diagnostic depth to the stakes of the decision it drives.
- Good-enough holds when evidence is clear, the part is the culprit, nothing upstream hunts the replacement, and the repair is recoverable.
- Go deeper on contradictions, victim parts, expensive actions, and callbacks.
- Watch your own drift: paralysis or false confidence.
- Document what you confirmed and what you reasoned but did not prove.
References
- Trade-standard practice on proportional diagnosis and confirmation before major repair
- OSHA and NFPA guidance requiring verification of safe conditions before clearing hazard-related faults
- See related: When the Cheap Fix and the Right Fix Differ (decision tree); Analysis Paralysis: When to Just Decide (decision tree); Documenting an Unsolved Fault for the Next Tech (how-to)