Customer Attempted a Repair Before the Call: Undo vs Build On Decision Tree
Why this matters
You arrive to find the customer already had a go at it. Maybe they swapped a part, maybe they rewired something, maybe they just took the cover off and "checked." Their attempt sits between you and the original fault. You face a fork: do you undo their work back to a known baseline before diagnosing, or do you accept their changes and build forward from where the system now stands?
Get this wrong in either direction and you waste time or create risk. Build on top of a botched DIY repair and you may chase a fault the customer created, never reaching the original problem. Undo good-faith work that happened to be correct and you charge for labor that fixed nothing. Worse, a customer modification can be unsafe in a way that is invisible until energized or pressurized. The decision tree below keeps you from diagnosing on top of an unknown.
This applies across every trade. A homeowner who jumpered a furnace control, a customer who swapped a faucet cartridge, someone who back-stabbed a receptacle, a person who replaced a dryer thermal fuse all leave you the same puzzle.
Symptom presentation
The tell is a system that is no longer in its as-built state. You see mismatched parts, fresh tool marks, wire nuts where there were none, tape over a connection, a fitting that does not match the others, or a component that is newer than its neighbors. The customer will often tell you what they did, but their account is incomplete by nature: they report what they intended, not every side effect.
The dangerous variant is the silent modification. The customer says "I didn't touch anything," but the evidence says otherwise. Treat any inconsistency between the story and the hardware as a reason to baseline before diagnosing.
Quick checks
- Is the attempted repair in the same subsystem as the complaint, or elsewhere? Same subsystem means it sits directly in your diagnostic path.
- Is the modification safe to energize or pressurize as it stands? If you cannot answer yes, do not power or pressurize until you correct it.
- Did the attempt introduce a new part, and is that part correct for the application? Wrong-rated parts are a frequent silent fault.
- Is the original complaint still present, or has it changed since the attempt? A changed complaint means the DIY work altered the fault picture.
Isolation tree
Step 1: Safety gate first. Inspect the modification for any condition that makes energizing or pressurizing unsafe: an undersized conductor, a missing ground, a connection without strain relief, a fitting without proper sealing, a defeated safety device. If found, you must correct it to a safe state before any further testing. This is non-negotiable and precedes all diagnostic logic.
Step 2: Determine path relevance. If the attempted repair is in a separate subsystem from the complaint and is safe, you may leave it and diagnose the original fault directly. Note it, but do not undo what is irrelevant and safe.
Step 3: For in-path modifications, decide undo versus accept. Undo to baseline when the attempt is unverifiable, when it introduces an unknown variable you cannot measure around, or when the parts are questionable. Accept and build on when the attempt is verifiably correct: a properly installed, correctly rated part, a sound connection, work you can confirm by measurement.
Step 4: When in doubt, baseline. The cost of restoring a known starting point is almost always less than the cost of diagnosing through an unknown. A system you understand from a clean baseline is faster to fault-find than a system carrying an undocumented change.
Step 5: Re-run the original complaint test from the baseline. Confirm whether the fault that prompted the call is still present once the system is in a known state.
Confirming diagnosis
You have correctly handled the prior work when the system is in a state you can fully describe and measure, every safety concern from the modification is resolved, and the original complaint is reproduced from that known baseline. ISO 13379-1 diagnostic logic depends on a defined reference state; a customer modification you neither measured nor reverted destroys that reference, which is precisely why baselining is the conservative default.
If you accepted the customer's work, you must be able to point to the measurement that verified it. "It looked fine" is not verification.
A customer's attempted repair can hide a life-safety defect: a defeated limit switch, a bonded neutral, a gas connection without a leak check, a back-fed circuit. OSHA 29 CFR 1910 and NFPA 70E require you to establish a safe condition before working on or energizing a system. Never assume a customer modification is safe because it powered on for them. Verify it yourself.
Next steps
If you undid the work: explain to the customer in plain terms why a known baseline was necessary, and bill the baseline labor transparently as part of reaching the diagnosis.
If you built on it: document which of their changes you verified and how, so there is no later confusion about what you touched versus what they did.
Either way, photograph the as-found modification before you change anything. It protects you, and it is the only record of what the system looked like when you walked in.
References
- OSHA 29 CFR 1910 General Industry standards, establishing a safe condition before energizing or working on equipment.
- NFPA 70E, Standard for Electrical Safety in the Workplace, establishing an electrically safe work condition.
- ISO 13379-1, Condition monitoring and diagnostics of machines, reference-state-based diagnosis.
- PHCC plumbing service guidance on verifying prior modifications before continuing work.