A Translated Symptom Description Doesn't Match What You Find: Decision Tree

Why this matters

You get a translated symptom, "it makes a loud banging noise," and you arrive to find a unit running dead quiet, with a completely different fault sitting in plain sight. Do you trust the translation, trust your eyes, or assume the customer is wrong. The mismatch is not automatically anyone's fault. It can come from a bad translation, an intermittent fault that is not present right now, a second unrelated problem, or the customer describing a symptom from days ago that has since changed. Sorting out which one you are dealing with, before you start replacing parts, is the whole skill.

Start here: is what you found capable of producing the described symptom

Before you second-guess the translation, ask the mechanical question first. If the fault you found genuinely could not produce the symptom described (a purely electronic error code cannot cause "banging"), the translation or the description is off. If the fault you found plausibly could produce something like the described symptom under different conditions (only under load, only at a certain point in a cycle, only when cold), you may have the right fault and just have not reproduced the exact condition yet.

If the fault could explain it, but only sometimes

1. Try to reproduce the reported condition, not just run the unit at idle. Run it under load, run a full cycle, wait for it to reach operating temperature. Many symptoms are condition-dependent and vanish the moment a tech shows up and the unit behaves for the audience.
2. If it reproduces under the right condition, you likely have the right fault; the translation was probably fine and you just needed to match the scenario.
3. If it still will not reproduce, treat it as intermittent and move to gathering more history rather than condemning a part on a symptom you have not personally witnessed. See the companion article on local fault history for a log-based alternative to a verbal description.

If the fault absolutely could not have caused the described symptom

1. Suspect the translation before the customer. Re-ask the same question through a different channel: instead of the app, try a gesture, a mimicked sound, or a written word, and see if the answer changes. A retranslation that comes back different confirms the first pass lost something.
2. Suspect a second, separate problem. Two independent faults on one job are common and get compressed into one confused description when the words are already thin. Ask, through your best channel, whether there is more than one thing wrong, or whether the noise happens in a different room, on a different fixture, or at a different time than what you are currently looking at.
3. Suspect a timeline gap. Ask specifically whether the described symptom is happening right now, or happened before and something has changed since. A translated past tense is easy to lose, and "it was banging, now it just doesn't turn on" is a completely different diagnostic path than "it is currently banging."
4. If none of the above resolves it, widen your physical search rather than trusting either the translation or your first find in isolation. Walk the whole system, not just the one component the description pointed at.

Do not silently override the customer

The temptation, once you have found something, is to fix what you found and never resolve why it did not match the story. That risks a callback for the actual reported problem, which is still sitting there. Close the loop: confirm back, through your best channel, that what you found and fixed matches what they described, or explicitly tell them you found and fixed a different issue and ask if the original complaint is still present.

Recap

1. Ask first whether the fault you found could plausibly produce the reported symptom under some condition.
2. Try to reproduce the exact reported condition before dismissing the description.
3. If it truly does not fit, re-check the translation through a second channel, consider a second unrelated fault, and check for a timeline gap.
4. Widen the search rather than picking a side between the words and your first find.
5. Confirm back explicitly whether the original complaint is resolved, not just whether you fixed something.

References

• Trade-standard practice for fault reproduction and intermittent-fault documentation.
• Manufacturer documentation on load-dependent and cycle-dependent fault conditions.
• See related: The Customer and Tech Don't Share a Language: Decision Tree; A Unit's Own Local Fault History.