How to check a corrective action worked

A corrective action can be complete and still not work. The new procedure is issued, the training is signed off, the record is closed, and three months later the same nonconformity turns up in a different shift. ISO 9001 clause 10.2 asks you to review the effectiveness of corrective action taken, and certification auditors often test it by sampling closed actions and looking for recurrence. This guide covers how to do that review so that it means something.

Implementation is not effectiveness

There are two separate questions to answer once an action has been agreed:

1. Was the action implemented? The register was changed, the form was revised and issued, the labels were moved. This can be checked as soon as the owner says the work is done.
2. Was the action effective? The cause has been removed and the problem has not recurred. This can only be checked after the process has run for a while under the new arrangements.

Many remediation processes stop at the first question. The record shows a revised document and a training sheet, the auditor ticks it, and nobody looks again. An effectiveness review is the second, later check, and it is the one that tells you whether the root cause analysis was right.

Keep the two checks as separate steps with separate dates. A corrective action that is implemented but not yet reviewed is not finished.

Define success before you start

The best time to decide how effectiveness will be judged is when the action is agreed, not when the check is due. Write down three things on the action record:

• The measure. What would show the problem has gone? A count, a rate, a sample result, an absence of something.
• The method. How it will be checked: record sample, observation, re-audit, data review.
• The date. When there will be enough data to judge.

Some examples:

Nonconformity	Measure of effectiveness	Method	When
Gauge used past its calibration date	No gauge in use past its due date; all transferred gauges have due dates	Sample 20 gauges on both lines; check every transfer since the change	3 months after the register change
Leavers keeping system access	All leavers disabled on their leaving date	Compare HR leaver list against account status for the quarter	End of next quarter
Supplier evaluations overdue	All active suppliers evaluated within 12 months	Review approved supplier list against evaluation records	After the next annual cycle
Wrong drawing revision on shop floor	No superseded drawings at workstations	Unannounced spot-check of 10 workstations	6 weeks after the new release process

Writing the measure down early also tests the action. If nobody can say what success would look like, the action is probably too vague.

What counts as evidence

Effectiveness evidence should be objective and should come from the process itself, not from the action owner’s confirmation. Strong audit evidence for effectiveness includes:

• Records generated after the change, sampled across the period: inspection records, account reports, evaluation forms.
• Trend data showing the problem rate fell and stayed down: complaints, returns, defects, incidents.
• Direct observation that people are working to the new method, ideally unannounced and on more than one shift.
• Interviews with the people doing the work, confirming they know and use the change.
• Absence of recurrence across a meaningful sample, stated with the sample size.

Weak evidence includes a signed training register on its own, an email saying “done”, a revised procedure with no sign that anyone uses it, or a check made the day after the change. These show implementation at best.

Attach the evidence to the action, not just a sentence saying it was reviewed. A certification auditor who samples the record should be able to see what was checked and what was found.

Grading the result

Three outcomes cover almost every case, plus a holding state while the check is pending:

• Effective. The measure was met across the sample, the cause is no longer present and nothing has recurred. Close the action.
• Partially effective. The problem has reduced but not gone, or the action worked in one area but not another. For example, line 2 gauges are all in date, but a transferred gauge on line 1 still lacked a due date because the old form was in use there. Keep the action open, add tasks for the gap and set a new check date.
• Ineffective. The problem has recurred at a similar rate, or the cause is still present. Reopen the investigation.
• Pending. The action has been implemented and the check date has not yet arrived.

Record the grade, the evidence and the name of the reviewer. Effectiveness in Orisynx is recorded on the action as effective, partially effective, ineffective or pending, so actions still waiting for their check stay visible.

Choosing how long to wait

The check date is a judgement about how much data will show recurrence. Too early, and an ineffective action looks effective simply because the conditions that caused the problem have not come round again. Too late, and the action has drifted out of anyone’s attention, while the problem may already be back.

Some factors to weigh when setting the date:

• How often the process runs. A gauge used every shift produces evidence within weeks. An annual supplier review produces one data point a year, so the effectiveness check may have to sit with the next cycle, with an interim check on the new method in between.
• How often the problem used to occur. If the nonconformity happened roughly once a quarter, a one-month check proves very little. Wait long enough that you would reasonably expect to see it again if nothing had changed.
• The conditions that triggered it. If the problem only appeared at month end, on the night shift or during a peak season, make sure the check period covers those conditions.
• The severity. For a high or critical finding, do an early interim check that the action is in use, then the full effectiveness check later. That way a weak action is caught sooner without pretending the early check proves effectiveness.

Record the reason for the chosen date alongside it. “Three months, covering at least six gauge transfers and one quarter-end” tells the reviewer what to look for and tells an auditor why that date was chosen.

When the action did not work

An ineffective result is useful information, not a failure of the person who carried out the check. It almost always means one of three things:

1. The root cause was wrong. The team stopped at a symptom, often “human error” or “lack of training”.
2. The cause was right but the action was weak. It depended on people remembering, such as a reminder email, when the cause needed a change to the system.
3. The action was not fully implemented. It reached one site, shift or system but not the others.

Reopen the original record rather than raising a new, unconnected one. The history matters: an auditor who sees the same nonconformity raised and closed three times without a link will reasonably conclude that the corrective action process is not working, and some certification bodies treat a repeat of a previous minor as grounds for a major. See major vs minor nonconformity for how that escalation usually works.

Go back to the investigation with the new evidence. Ask what the recurrence shows that the first analysis missed. The guide on root cause and corrective action covers methods for that second look.

Building checks into the programme

Effectiveness checks are easy to forget because they fall due months after the finding, when everyone has moved on. A few habits keep them on track:

• Put the check date on the action when it is agreed, with a named reviewer.
• Combine checks with scheduled audits where the timing fits. At Kestrel Metals, a fictional fabrication business, the September welding audit was scoped to check the effectiveness of every action from February’s audit of the same line.
• Use a follow-up audit for high-severity findings where waiting for the next routine audit is too long.
• Report on overdue checks at management review. A list of actions that have been implemented but never reviewed is a useful signal of where the process is slipping.
• Sample closed actions yourself before the certification audit. Pick five at random and ask whether the evidence would convince an outsider.

At Kestrel Metals, R. Patel reviews open effectiveness checks monthly with the auditors, J. Okafor and K. Lewis. Each closed action has a one-line conclusion and the evidence attached, for example: “Effective. 20 of 20 sampled gauges in date across both lines; all 6 transfers since July carry due dates. Reviewed by J. Okafor, 12 January.”

That line is what a certification auditor hopes to find, and it is also the clearest proof to the business that the time spent on corrective action was worth it.