Findings are normal. Every inspecting organisation gets them. What actually separates sites at review time isn't whether they've ever had a finding — it's what their record shows happened after the last one. A corrective action that genuinely closed builds trust; one that reappears destroys it. This expands on the closing section of the QMS audit pillar guide.
The three parts of a closed corrective action
A corrective action that holds up has three components, and skipping any one is why findings come back:
- The instance fixed. The specific record corrected, the expired calibration renewed, the missing signature obtained. This is the part everyone does — and the part that, on its own, isn't enough.
- The cause addressed. Why did it happen, and what changed so it can't simply recur? An expired calibration fixed by recalibrating is an instance fix; an expired calibration fixed by putting the renewal on a register that surfaces it in advance is a cause fix. Only the second one prevents the next occurrence.
- The evidence kept. The finding, the action taken, and the verification that it worked — recorded in the improvement record, so the next review sees a closed loop rather than a silent gap or a repeat.
Miss part two and you've treated a symptom. Miss part three and you can't prove you treated anything.
Why instance-only fixes feel like progress but aren't
Instance fixes are seductive because they're fast and visible. The reviewer noted a blank checksheet field; you fill in the process and move on. But nothing about why the field was blank changed — the same time pressure, the same paper form, the same reliance on memory is still there, and next quarter it produces the same blank somewhere else. The finding didn't get fixed; it got postponed.
Cause analysis doesn't have to be elaborate. Usually one honest "why does this keep being possible?" gets you there: because the record depends on a person remembering under pressure. The cause fix is almost always structural — change the system so the failure can't occur, rather than asking people to try harder.
The improvement record is the proof
The QMS improvement record is where this lives, and reviewers read it closely because it's the clearest signal of a healthy quality system. They look for findings entered, actions with causes addressed, and verification that the fix held. A thin improvement record with only auditor-found items suggests a site that waits to be told; a fuller one with self-found items, each closed with cause and evidence, suggests a site that runs itself.
The worst pattern is the recurring finding — the same issue at consecutive reviews. It doesn't just count against that item; it undermines confidence in every other corrective action on the register, because it demonstrates the site's "closed" doesn't mean closed.
Closing the loop, by design
The reliable way to make corrective actions stick is to (a) make most of them cause fixes and (b) make the loop impossible to lose track of. A system that holds each finding with its action, owner, due date and verification — and won't let an item sit "open" silently — turns the improvement record from a list into a managed process. That's part of what the QMS Compliance Centre is built to do; and because so many cause fixes amount to "stop relying on memory", the same digital checksheets and live registers that prevent the findings also are the corrective actions.
The goal is simple to state and hard to fake: a reviewer reading your improvement record should conclude that when this site finds a problem, it fixes the reason — and the record proves it every time.