How to Test Corrective Action Effectiveness in 30 Days

A corrective action can look closed while the original risk remains alive. The form has a responsible person, a due date, a photograph, and a green status. The worksite, however, may still be teaching the same shortcut, tolerating the same weak barrier, or depending on the same supervisor memory that failed before the incident.

This is why incident investigation should not end at action closure. It should end only after the team has tested whether the action changed the condition, behavior, decision, or control that allowed the event to happen. Across 25+ years in executive EHS roles, Andreza Araujo has seen corrective action systems fail less because people refuse to act and more because they confuse completion with risk reduction.

The 30-day review below is designed for EHS managers, site leaders, and supervisors who need a practical way to test effectiveness after an incident. It connects James Reason's distinction between visible acts and latent conditions with Andreza Araujo's field thesis in Safety Culture: From Theory to Practice: culture becomes visible when leaders verify whether the system changed, not only whether the document was signed.

Step 1: Freeze the original risk statement

Before reviewing any corrective action, freeze the original risk statement in one sentence. The sentence should describe the unwanted event, the failed control, and the exposure that remains if nothing changes. If the team cannot write that sentence, it is not ready to judge effectiveness.

A weak statement says, "operator did not follow procedure." A useful statement says, "the line-break permit allowed maintenance to open a pressurized connection because isolation verification was visual, informal, and not witnessed by the supervisor." The second version points toward a control that can be tested.

This first step protects the review from drift. Corrective actions often expand into training, reminders, and new forms because those outputs are easy to close. The frozen risk statement keeps the team asking whether the specific exposure was reduced. It also links the review to the incident timeline built in the first 24 hours, where the sequence of decisions should already be visible.

Step 2: Separate containment from correction

The first 30 days after an incident contain two kinds of work. Containment prevents immediate recurrence while the investigation is still open. Correction changes the system so the same exposure is less likely to return. When teams mix the two, they close emergency patches as if they were permanent fixes.

Locking out a machine, pausing a contractor crew, or adding a temporary supervisor check may be necessary containment. It does not prove the root condition has changed. A corrective action should survive after the emergency attention fades, after the manager stops asking daily questions, and after the crew returns to normal workload.

ISO 45001:2018 requires organizations to react to incidents and determine action to eliminate causes, when applicable. That requirement matters because a temporary pause is not the same as eliminating a cause. The review team should label each action as containment, correction, or verification so leaders do not confuse speed with depth.

Step 3: Choose the proof before inspecting the action

Effectiveness testing fails when the reviewer starts with the action file. The file usually contains evidence that the task was completed, not evidence that the risk changed. Instead, choose the proof first.

For a procedural weakness, proof may be three sampled permits showing the new verification step used correctly. For a supervision gap, proof may be field observation across different shifts. For a design weakness, proof may be a physical modification, commissioning record, and operator demonstration. For a reporting weakness, proof may be better near-miss quality, not a higher number alone.

Andreza Araujo's work across more than 250 cultural transformation projects shows that leaders often overvalue documentary proof because it is comfortable. Field proof is less polite. It shows whether the person closest to the risk can explain the change, use the control, and challenge the task when the condition is not met.

Step 4: Test the barrier, not the memory

Many corrective actions depend on memory. The worker was retrained, the supervisor was reminded, the contractor received a toolbox talk, or the rule was reissued. These actions may be useful, but they are weak if the original event involved a missing physical barrier, unclear authority, poor design, or an incentive to hurry.

The 30-day review should ask which barrier now prevents recurrence when memory fails. If the answer is only "people know better," the action has probably not changed the system enough. James Reason's work on organizational accidents is useful here because it pushes the reviewer to look beyond the active failure and ask which latent conditions allowed the event to make sense at the time.

For high-potential incidents, connect the review to barrier analysis before RCA. A corrective action is effective only if the relevant barrier is stronger, more visible, easier to use, or harder to bypass than it was before.

Step 5: Interview the people who must live with the fix

The action owner can tell you what was done. The crew can tell you whether it works. Interview at least three people who must live with the corrective action, preferably across different shifts, roles, or contractor groups.

The questions should be concrete. Ask what changed in the job, when the new control is used, what makes it difficult, who can stop the task if the control is missing, and what would happen during a rushed restart. Avoid asking whether the action is good. That question invites politeness. Ask instead how the job is different.

This is where safety culture enters the review. In The Illusion of Compliance, Andreza Araujo warns against systems that look complete while field behavior remains unchanged. If the crew cannot describe the fix without reading the procedure, the action may have entered the system without entering the work.

Step 6: Sample the next three similar jobs

Effectiveness cannot be judged from the incident job alone. Sample the next three similar jobs, tasks, permits, shifts, or contractor activities where the same exposure could appear. The number three is not a statistical claim. It is a practical floor that prevents a single staged observation from becoming false confidence.

If the incident involved lifting, review the next three lifting plans. If it involved chemical transfer, observe three transfers. If it involved a confined space entry, inspect three entry preparations. The point is to see whether the corrective action travels from the investigation room into normal work.

This step also protects against local fixes that never reach adjacent work. A site may correct one crew and leave another crew exposed because the action was assigned to the person closest to the event rather than to the owner of the process. The post-incident action plan should define where replication is required.

Step 7: Check whether the metric changed for the right reason

Metrics can help the 30-day review, but they can also mislead it. Fewer deviations may mean the control improved. They may also mean supervisors stopped recording deviations because the incident made everyone nervous. More near misses may mean risk increased, or they may mean reporting trust improved.

The review should pair a lagging or leading number with field evidence. If permit quality improved, sample the permits and ask workers what changed. If corrective action closure time improved, test whether actions closed with proof or only with uploaded documents. If observation findings fell, verify whether observers are still entering the field.

This is why the action review should link to control effectiveness metrics without becoming a spreadsheet exercise. The metric is useful when it points leaders back to the control. It is weak when it replaces inspection.

Step 8: Decide, document, and reopen when needed

At day 30, the review team should make one of three decisions. The action is effective, the action is partially effective and needs adjustment, or the action is ineffective and must be reopened. Avoid the fourth unofficial decision, which is leaving the action green because reopening it would embarrass the owner.

The decision should include the proof used, the people interviewed, the sampled jobs, and the residual risk. If the action is reopened, the record should explain what failed in the fix, not who failed to make the dashboard look clean. That distinction keeps the review technical rather than political.

Andreza Araujo's experience in multinational operations, including the PepsiCo South America period where the accident ratio fell 50% in 6 months, reinforces a plain lesson: leaders get better safety performance when they discipline the management routine around evidence. Closure is a promise. Effectiveness review is where the organization checks whether the promise reached the work.

FAQ

What is corrective action effectiveness after an incident? Corrective action effectiveness is the verified evidence that an action reduced the exposure identified by the investigation. It is different from action closure because closure proves that a task was completed, while effectiveness proves that the risk condition changed.

When should an effectiveness review happen? A 30-day review works well for many operational incidents because it gives the site time to apply the fix in real work without letting the action disappear into the management system. High-severity risks may need earlier checks and repeated verification at 60 or 90 days.

Who should test corrective action effectiveness? The test should include EHS, the action owner, the supervisor responsible for the work, and people who perform the task. For high-potential events, a manager outside the area should review the evidence so the decision is not only local self-confirmation.

What evidence proves a corrective action worked? Useful evidence includes field observation, sampled permits or records, worker interviews, physical barrier verification, control demonstrations, and trend data interpreted with context. A photograph or attendance sheet may support the file, but it rarely proves risk reduction by itself.

Should ineffective actions be reopened? Yes. Reopening an action is a sign that the organization is protecting risk evidence, not a sign that the first owner failed. If leaders punish reopened actions, teams will learn to protect the dashboard instead of protecting people.

Final check for the EHS manager

A corrective action effectiveness review should answer one question in plain language: can the same exposure return tomorrow under normal production pressure? If the answer is yes, the action is not finished, no matter how complete the file looks.

For deeper work on the cultural side of verification, start with Safety Culture: From Theory to Practice and The Illusion of Compliance. To bring this review discipline into a leadership routine, see Andreza Araujo's safety culture and EHS transformation work.