The CAPA that closes the loop: containment, root cause and proof it worked
A corrective action that is never checked for effectiveness was never actually proven to work. How to separate containment from root cause, name the method behind it, and catch a defect before it recurs a third time.
The totalMOS team5 min read
Key takeaways
- Containment, correction, root cause and prevention are four different jobs; one sentence that tries to cover all four usually only does the first.
- A root cause with no named method, such as the five whys or a fishbone diagram, is a guess with a deadline attached.
- A CAPA is not closed until someone checks the actual readings, weeks later, to see whether the defect stopped.
- The same defect type opening a second non-conformity within weeks is the plant’s own evidence that the last CAPA didn’t address the real cause.
Four jobs hiding in one text box
Most corrective action forms have one box for what happened and what was done about it. Filling it in asks a technician to do four different jobs at once: stop the bad product that already exists, fix what is wrong with the machine or the process, explain why it failed, and change something so it cannot fail the same way again. Squeezed into one sentence, the easiest of the four wins, and it is never the last one.
- Containment: what happens to the product already made, put on hold, segregated, sampled.
- Correction: what is done to the machine, the process or the person, right now, to get the line running again.
- Root cause: why the failure happened, not what failed.
- Prevention: what changes so the same root cause cannot produce the same failure again, on this machine or another one.
A worn piston seal gets replaced and the line restarts: that is a correction. It says nothing about why the seal failed at six weeks when the spec calls for twelve, and nothing changes in the replacement schedule, the spare-parts count or the supplier to stop seal number four from failing the same way.
A root cause needs a named method
“Operator error” and “machine issue” are not root causes; they are where the investigation stopped. A root cause answer keeps asking why past the first plausible answer, in writing, so the next person reading the record can follow the reasoning instead of taking it on faith.
Two methods cover most of what shows up on a packaging line. The five whys asks why the failure happened, then why that happened, usually three to five times, until the answer names something that can actually be changed. A fishbone diagram earns its place when more than one cause looks plausible at once, a worn part, a new supplier lot, a humid week, by sorting the candidates into people, machine, material, method and environment before choosing which one to chase.
An owner, a due date, and what was actually done
- 1Name one person responsible for each action, not a department or a shift.
- 2Give the action a due date, never “ongoing”.
- 3Separate the corrective action, which can often close in days, from the preventive action, which may need a purchase order, a schedule change or training that takes weeks.
- 4Record what was actually done, with the date, not what was planned. A fix that was approved but not yet installed is not a closed CAPA.
The step almost everyone skips
A corrective action gets written, signed and filed, and that is usually where it stops. Nobody comes back in a few weeks to check whether the defect actually went away. That return visit, the effectiveness verification, is the only thing that turns “we did something” into “it worked”.
It does not need to be elaborate. Pull the readings for that test, on that machine, for the weeks since the fix went in. If the defect type has not come back, write that down and close the CAPA. If it has, the CAPA reopens and the investigation goes back to root cause, because the first answer was wrong or incomplete.
Catching recurrence before the third time
A CAPA that did not take rarely announces itself. It shows up as the same defect type, on the same machine, two or three non-conformities apart, each one investigated on its own with no reference to the last.
A Pareto by defect type, read over weeks rather than one shift, is what surfaces it. A bar that keeps reappearing after a corrective action was supposedly closed against it is the plant’s own evidence, sitting in its own data, that the fix did not reach the real cause.
How totalMOS handles it
In totalMOS, a CAPA is a structured record rather than a free-text box: containment, a root cause with its method named, the corrective action, the preventive action, an owner, a due date and an effectiveness verification, all attached to the same non-conformity as its hold and disposition, with the audit trail underneath. The Pareto by check and by machine carries recurrence across CAPAs instead of leaving each one to be investigated as if it were the first. The weight curve and the audit-ready record are two of the places a CAPA usually starts; the cosmetics demo plant shows one followed to its root cause.
Frequently asked questions
What is a CAPA?
Corrective and preventive action: the record of what fixed a specific non-conformity and what changed so the same root cause cannot produce it again, with a named owner, a due date, and a check, after the fact, that it worked.
What is the difference between corrective and preventive action?
Corrective action fixes the failure in front of you, such as replacing a worn part. Preventive action changes something structural, a replacement schedule, a spec, a supplier, so the same root cause cannot produce the same failure on a different machine or a different day.
How do you verify a CAPA was effective?
By checking the actual readings for that test, on that machine, over the weeks after the fix, not by reviewing the paperwork. If the defect type has not reappeared, the CAPA can close; if it has, the investigation reopens.
