It Worked Yesterday: What Changed

Why this matters

When a system runs fine for years and then quits overnight, the cause is rarely slow wear - wear is gradual and gives warning. A sudden failure almost always traces to something that changed: an event, an action, a condition that flipped. Find what changed and you find the fault. Chasing it as if it were a long-developing problem sends you down the wrong path and wastes the visit.

Start here: anchor on the change

The complaint itself is your best clue: "it worked yesterday." That means something is different today. Your whole diagnosis is a hunt for the difference. Resist the urge to start pulling parts. First ask: what is not the same as when it worked?

Interview the customer with that single goal. The answer is usually in their first two sentences if you listen for a change rather than a symptom.

Step 1: was there a triggering event

Sudden failures usually have a trigger. Ask directly about each:

• Power or supply event. A storm, an outage, a surge, a tripped breaker, a supply interruption, a shutoff that got closed. Anything that cut or spiked the feed.
• Weather change. A hard freeze, a heat spike, heavy rain, high wind. Temperature swings stress connections, fluids, and seals all at once.
• Physical event. Something bumped, dropped, moved, flooded, or struck near the equipment. A vehicle, a renovation, a leak overhead.
• A person touched it. Someone "adjusted" a setting, flipped a switch, unplugged something, or cleaned around it and knocked a connection loose. This is far more common than customers volunteer - ask plainly and without blame.

If any of these line up with the timing, that is your prime suspect. Investigate the trigger's likely damage first.

Step 2: did anything get serviced or changed nearby

• If recent work was done on this system or an adjacent one, the change may be a side effect. A repair next door that shut off a shared supply, a setting changed during unrelated service, a part disturbed during access. (See the related article on one fix breaking another thing.)
• If a setting or schedule changed, the system may be doing exactly what it was told - the instruction is the fault, not the hardware.

Step 3: did the load or usage change

• If demand jumped, a system that coped at light load can fail at heavy load. More users, more draw, a new device added to the circuit, a seasonal spike. The hardware did not change; the stress on it did.
• If usage stopped and restarted, a system left idle and then brought back can fail on restart even though it ran fine before the pause. (See the related article on seasonal first-startup failures.)

Step 4: nothing changed externally - now suspect the part

If no event, no service, no load change, and no human action explains the timing, you are in the rarer case of a genuinely sudden component failure. Some parts do fail abruptly with no warning - a snapped connection, a shorted device, a seized mechanism, a protective component that finally let go.

Now work it as a standard fault: compare readings against baseline, isolate the suspect, find the dead leg. But only reach this step after the change-hunt comes up empty, because the change-hunt is faster and right far more often.

The interview is the diagnosis

On a "worked yesterday" call, your most powerful tool is the conversation, not the meter. A sharp two-minute interview about what changed will out-diagnose an hour of random testing. Write down the timeline and what changed; it points the meter exactly where to go.

References

• Trade-standard root-cause and event-correlation practice
• Manufacturer specifications (for confirming a suspected sudden failure)
• See related: Fixed One Thing, Broke Another
• See related: Seasonal First-Startup Failure Decision Tree