I have reviewed thousands of corrective actions in my career. 8D reports, 5-why analyses, fishbone diagrams, Ishikawa exercises, CAPA systems with elaborate workflows and automated tracking. And after all of that, I can tell you with confidence that approximately eighty percent of the corrective actions I see are not corrective at all. They are containment dressed up as correction. Band-Aids on bullet wounds.
The problem is not that people are lazy or incompetent. The problem is that genuine corrective action is hard, expensive, and uncomfortable. And most quality systems are designed to close findings, not to solve problems. So the path of least resistance is to write a corrective action that looks credible, get it approved, and move on.
The retraining fallacy
If I had to name the single most common corrective action I have seen in twenty years of quality work, it would be this: "Retrained operator on the correct procedure."
This is not a corrective action. It is an admission of failure disguised as one. If an operator made a mistake, retraining addresses the symptom — the operator did not know or did not follow the procedure. It does not address the root cause: why was it possible for the operator to make the mistake in the first place?
Was the procedure clear? Was it accessible at the point of use? Was the workstation designed to make the correct action easier than the incorrect action? Was the tooling adequate? Was the scheduling putting the operator under time pressure that made shortcuts rational? These are the system conditions that produced the error. Retraining the operator does not change any of them.
If your corrective action does not change the system, the system will produce the same defect again. The only question is whether it happens on your watch or the next guy's.
The five-why that stops too early
The 5-why analysis is the most abused problem-solving tool in quality. Not because the method is flawed — it is fundamentally sound. But because people stop at the level of cause that is comfortable to fix.
I reviewed a 5-why recently for a machining defect that reached a customer. Here is what the team wrote:
Why 1: The operator did not check the fixture before running the batch.
Why 2: The fixture check was not listed in the work instruction.
Why 3: The work instruction was last updated two years ago.
Why 4: The engineering team did not have a review schedule for work instructions.
Why 5: There is no process for periodic review of work instructions.
Corrective action: "Implemented a fixture check step in the work instruction."
Do you see what happened? They identified a systemic gap — no process for reviewing work instructions — and then fixed a single instance of it. The corrective action addresses Why 2 but ignores Why 5. In six months, a different work instruction will be outdated, a different operator will miss a different step, and a different defect will reach a different customer. The system was never fixed. The bullet wound was dressed with a Band-Aid.
What real corrective action looks like
Genuine corrective action has three characteristics that distinguish it from the pretender:
1. It changes the system, not the symptom. If you remove the corrective action, the defect should not recur because the conditions that produced it no longer exist. If removing your corrective action would allow the defect to return, you have not corrected anything. You have added a layer of inspection that catches the defect before it escapes.
2. It addresses the deepest root cause the organisation is willing to fix. Not every root cause can be eliminated. Some are constrained by cost, technology, or customer requirements. But the decision to accept a root cause should be a conscious, documented choice — not an accident of where the 5-why analysis happened to stop.
3. It is verified with data, not with a signature. I close corrective actions with a verification step: run the process for a defined period under defined conditions and measure the defect rate. If the defect rate has not dropped to the expected level, the corrective action failed. Reopen the 8D. The signature of the quality manager is not verification. Data is verification.
Why we settle for Band-Aids
The honest answer is that real corrective action takes time and money. It requires shutting down a process, engineering a fix, validating the fix, and monitoring it. It requires cross-functional collaboration between quality, engineering, production, and sometimes procurement. It requires a management team that is willing to accept short-term pain for long-term gain.
Band-Aids require none of these things. They require a quality engineer with a template and a deadline. And since the certification auditor is checking whether corrective actions exist and are documented — not whether they actually worked — the system rewards documentation over effectiveness.
I once inherited a plant that had 340 open corrective actions in their system. Three hundred and forty. When I dug into them, I found that 270 were duplicates or near-duplicates — the same root cause, addressed with the same Band-Aid, failing and being reopened under a new number. The system was not correcting problems. It was cataloguing them.
We shut the system down. We took the top ten recurring failure modes, ran genuine root cause analyses, fixed the systems, and verified the results. We closed 270 of the 340 actions in three months. Not by writing better reports — by solving the actual problems.
Stop writing corrective actions that do not correct. Start fixing systems. It is harder, slower, and more expensive. But it is the only thing that actually works.
