It Was Fine Until We Changed Something: A Decision Tree

Why this matters

The most reliable diagnostic clue you will ever get is "it worked until X." Whatever happened right before the failure is your prime suspect, every time, in every trade. The mistake is treating the new symptom as a random breakdown and chasing it from scratch when the cause is sitting right there in the timeline. Change analysis is the fastest path to root cause and it costs nothing but a few good questions.

Start here: pin down the change

Before you open anything, get the story straight.

• What changed, and when exactly? A part replaced, a setting adjusted, a panel opened, a valve turned, an update applied, new equipment added, work done by someone else, weather that shifted, power that blinked.
• Did the symptom start immediately, or after a delay? Immediate failure points at the change directly. A delay of hours or days can still be the change (a slow leak, a loosened connection that finally let go), but it widens the field.
• Is the change reversible? If you can undo it and the symptom clears, you have your answer in one move. Always try this first when it is safe.

Write the timeline down. Half of these jobs solve themselves the moment the sequence is clear.

Branch 1: a part was replaced or serviced

New parts fail diagnostics in predictable ways.

• Wrong part. Right shape, wrong spec. A capacitor of the wrong rating, a valve with the wrong orifice, a fuse or breaker of the wrong size, a filter that restricts more than the original. If the replacement does not match the nameplate or the original spec, that is your fault until proven otherwise.
• Right part, installed wrong. Reversed polarity, a line on the wrong port, a connector half-seated, a wire landed on the wrong terminal, a gasket pinched. The part is good; the install is not.
• Something disturbed during the work. A neighboring connection bumped loose, a wire nicked, debris dropped into a line, a setting knocked off. The repair was fine but the surrounding system got jostled.

Branch 2: a setting was adjusted

Settings changes are easy to miss because nothing physically moved.

• If a control, switch, or dial was touched, return it to its prior value first. Thermostats, pressure switches, flow rates, timers, limit settings, dip switches, software config.
• If someone "optimized" something, be suspicious. Cranking a setting to its limit, disabling a feature, overriding a safety. The system was balanced; the change unbalanced it.
• If a mode was switched, confirm it matches the season and the need. Heat versus cool, manual versus auto, a bypass left engaged after the last job.

Branch 3: an external change

Sometimes nobody touched the equipment but the world around it moved.

• Power event. A surge, an outage, a brownout. After a power blink, equipment can come back in a fault state, lose a setting, trip a safety, or fail a component that was already weak. If power was interrupted, check for tripped protection and lost configuration before anything else.
• Supply change. Water pressure shifted, gas service worked on, voltage changed, a shared circuit got a new load. The equipment is fine; what feeds it changed.
• Load or environment change. New construction nearby, a remodel, a new appliance on the same system, a seasonal swing. The system that handled the old conditions is now seeing different ones.

Branch 4: it was coincidence

Occasionally the change and the failure really are unrelated. Treat this as the last resort, not the first. Only accept coincidence after you have undone the change (or proven it is good) and the symptom persists unchanged. The bar is high because change-correlation is right far more often than it is wrong.

Working the tree cleanly

Undo or verify the suspected change before introducing any new variables. Change one thing at a time so you always know what fixed it. If you must keep the change, prove independently that it is correct rather than assuming it. Document what changed and what you found, because the same person who changed it once will change it again.

References

• See related: Symptom Here, Cause Elsewhere: A Decision Tree
• See related: The Power Supply vs the Load: A Decision Tree
• Trade-standard troubleshooting practice (change analysis as first step to root cause)
• Manufacturer service documentation for correct settings, part specs, and reset procedures