Two Symptoms, Which Is the Cause

Why this matters

You arrive and find two things wrong. Fix the wrong one and the other keeps coming back, and now you have a callback. Most multi-symptom calls are one root cause throwing two shadows, or one fault that triggered a second. The skill is separating cause from effect before you spend parts and labor on a symptom. Get the order right and you fix it once.

Start here: are these two faults or one

Before deciding which is the cause, decide whether they are even separate. Three relationships are possible:

• One cause, two symptoms. A single fault produces both complaints. Fix the cause and both clear.
• Cause and effect chain. Symptom A caused symptom B. Fix A and B stops; fix B alone and it returns because A is still live.
• Two unrelated faults that happen to coexist. Both need fixing independently.

Your whole diagnosis is figuring out which of these three you have, and you do it before pulling anything.

Step 1: did they start together or one after the other

Timing is the strongest clue you can get cheaply.

• If both began at the same moment, suspect one common cause feeding both. A shared supply, a shared control, a single event that hit both. Look upstream for the one thing they have in common.
• If one clearly came first and the second followed, the first is your prime suspect for causing the second. Fix the earlier one and see if the later one resolves.
• If you cannot tell, interview for it. The customer's timeline is often the whole answer.

Step 2: do they share a common upstream

Trace each symptom back toward its source.

• If both trace to the same feeder, control, or component, that shared element is almost certainly the single cause. One bad supply can starve two devices; one bad control can misbehave in two ways. Test the shared element first - fixing it may clear both complaints at once.
• If they trace to completely separate paths with nothing in common, you likely have two independent faults. Diagnose and fix each on its own.

Step 3: can one symptom physically cause the other

Reason about the mechanism.

• If symptom A could plausibly produce symptom B (a leak causing a level drop, an over-current causing a part to fail, a restriction causing both low output and high stress), then A is the cause and B is the effect. Fixing A is the only durable repair.
• If neither can cause the other, they are either a common-cause pair (step 2) or unrelated (two faults).

Think it through before you act. A wrong cause-and-effect guess is what produces the repeat visit.

Step 4: test the hypothesis before committing

You do not have to guess blind. Often you can prove the relationship cheaply.

• Address the suspected root cause (or temporarily relieve it) and watch whether the second symptom changes. If clearing A makes B improve, you have confirmed the chain.
• If clearing the suspected cause does nothing to the second symptom, they are independent - revise and treat them separately.

The cardinal rule: never chase the symptom

The downstream symptom is the loud, obvious one - the dead device, the visible leak, the alarm. The cause is often quiet and upstream. The instinct to fix the loud thing is exactly the trap. Always ask "what made this happen" before "how do I replace this." Replacing a part that a live root cause will destroy again is the most expensive mistake in the trade.

When you have identified the true cause, fix it, then confirm both symptoms are gone by running the system through the failure condition. Two symptoms in, zero symptoms out.

References

• Trade-standard root-cause analysis practice
• See related: Callback: The Fix Didn't Hold
• See related: Fault Only Under Specific Conditions
• See related: Baseline vs Faulty Comparison Diagnosis Method