The Investigation Quality-Corrective Action Link: Why Your Fix Failed Before You Implemented It

When a corrective action fails to prevent recurrence, the first questions usually focus on implementation. Was the action completed properly? Was there enough ownership? Did supervisors reinforce the change? Was the new process embedded into normal work?

Those are reasonable questions, but they start too late.

In many cases, the corrective action was already compromised before anyone began implementing it. 

The problem occurred during the investigation, when the organisation gathered incomplete evidence, accepted assumptions as fact, stopped its analysis too close to the event, or failed to identify the organisational conditions that allowed the incident to develop.

By the time the corrective action reached the action register, the organisation was trying to solve the wrong problem.

That creates an important connection between investigation quality and corrective action sustainability. Corrective actions cannot be stronger than the findings they are built on. Findings cannot be stronger than the analysis behind them, and the analysis depends on the quality of the evidence gathered during the investigation.

In practical terms, the chain looks like this:

Evidence → Contributing Factors → Analysis → Findings → Corrective Actions → Risk Reduction

Weakness anywhere upstream affects everything that follows.

The corrective action is downstream of the investigation

Organisations often treat investigations and corrective actions as separate processes. The investigation team completes its report, recommendations are entered into an action management system, owners are assigned, due dates are established, and the organisation moves into implementation mode.

That separation can hide a fundamental problem. A corrective action is not an independent solution. It is the final product of the investigation's reasoning.

If the investigation correctly identifies an Absent or Failed Defence, understands the conditions that affected the work, and identifies the Organisational Factors that contributed to those conditions, there is a strong basis for developing corrective actions. We should specifically link corrective actions to the Absent or Failed Defences and Organisational Factors identified as contributing factors.

If that analysis is weak, the action may still look professional. It can have an owner, budget, due date,

implementation plan and verification process. None of those things prove that it addresses the conditions that actually contributed to the incident.

You can implement the wrong corrective action extremely well.

That is one reason repeated incidents can be so frustrating. From the action management perspective, the organisation may have done everything expected. Actions were assigned. Completion rates were high. Evidence of closure was supplied. Yet months later, a similar event occurs.

The natural response is to question whether the previous actions were properly implemented. Sometimes that is exactly the issue. But another possibility deserves equal attention: perhaps the original investigation never gave the organisation the information it needed to design an effective response.

Data quality sets the ceiling

The quality of an investigation depends heavily on what happens before the team starts filling in an ICAM chart.

Data gathering is not simply the administrative stage that precedes the analysis. It determines what the analysis can see.

The ICAM investigation process uses PEEPO to structure evidence gathering across People, Environment, Equipment, Procedures and Organisation. The quality and level of detail developed through that process are critical because contributing information identified during data gathering feeds the later ICAM analysis.

This matters because incidents rarely arrive with their contributing factors neatly labelled.

Consider an event where an experienced worker bypasses part of a procedure while completing a routine task. If the investigation gathers only the procedure, the worker's statement and evidence that the required step was missed, the conclusion may appear straightforward. The worker did not follow the procedure.

The resulting actions are predictable: retrain the worker, reissue the procedure, conduct a toolbox talk, remind the workforce of compliance expectations.

But what happens if the investigation gathers more complete evidence?

Perhaps the required equipment had been unavailable for several weeks. The written procedure no longer reflected the way the task was actually performed. Supervisors knew about the workaround and had seen it used previously.

Production targets made the formal method difficult to apply within the available time. A previous equipment change had altered the task, but the implications had not been fully addressed through the management of change process.

The worker's action has not disappeared from the investigation. It is still relevant evidence. What has changed is the explanation.

The first investigation asks, “Why didn't the worker follow the procedure?”

The stronger investigation asks what conditions made that behaviour possible, understandable or likely, and what controls should have prevented the event from developing.

Those questions lead to very different corrective actions.

Why investigations stop too close to the person

Stopping at the action of the person closest to the event is attractive because it produces a simple story. Someone made a decision, failed to follow a requirement or performed a task incorrectly. The organisation can identify that behaviour, assign an action and move on.

ICAM deliberately pushes the analysis further.

We describe the analysis as working back from the Absent or Failed Defences through Individual or Team Actions and Task or Environmental Conditions to the Organisational Factors that influenced the event. This approach helps prevent the investigation from being restricted to operator errors or violations and instead identifies the workplace and organisational conditions that contributed to what occurred.

That distinction has direct consequences for corrective action quality.

If the investigation concludes that a person made an error, the natural response is to try to change the person. Training, coaching, reminders and additional procedural requirements follow easily from that diagnosis.

If the investigation establishes that equipment design made an error difficult to detect, the corrective action conversation changes. If supervision arrangements did not provide the required verification for a high-risk task, the response changes again. If incompatible operational goals created pressure that routinely encouraged a workaround, the action needs to address that condition rather than simply tell people not to use the workaround.

The corrective action is shaped by where the investigation stops.

Organisational conditions determine whether the fix has somewhere to stick

Organisational Factors are sometimes treated as background context rather than part of the causal explanation. That weakens both the investigation and the actions that follow.

Within ICAM, Organisational Factors include management decisions, processes and practices that influence workplace conditions and the effectiveness of defences. They can involve areas such as procedures, design, communication, contractor management, maintenance management, management of change and organisational learning.

These conditions may exist well before the incident. They can remain unnoticed until they combine with other contributing factors and an event exposes the weakness.

This is why apparently sensible corrective actions sometimes fade after implementation.

Imagine an investigation identifies inconsistent pre-start checks and recommends retraining the workforce. Training is delivered and competency records are updated. For several weeks, compliance improves.

But the investigation never establishes why the checks had become inconsistent.

If supervisors are managing an unrealistic span of control, the new training does not change that condition. If the checklist does not reflect the equipment configuration, training does not correct it. If operational planning routinely compresses the time available for the check, training does not remove that pressure. If verification is absent, the organisation may not even know when the behaviour begins drifting again.

The action can be completed exactly as written and still leave the conditions that influenced the original event in place.

That is not primarily an implementation problem. It is a diagnosis problem.

Why “more training” appears so often

Training is one of the easiest corrective actions to recommend because it is familiar, measurable and relatively simple to close. Attendance can be recorded, competency assessments completed and evidence uploaded into the action system.

None of that makes training inherently weak. Where the evidence establishes a genuine competency gap, training may be entirely appropriate.

The problem begins when training is used as the default response to behaviour without establishing why that behaviour occurred.

Weak responses such as providing additional training or reminding supervisors of their responsibilities can miss the system conditions influencing supervisory performance. Stronger responses may require examination of supervision requirements for high-risk work, verification controls, supervisor workload, span of control, performance expectations or feedback mechanisms.

The question is therefore not, “Is training a good corrective action?”

The better question is, “What finding does this training action address?”

If the investigation cannot answer that clearly using evidence, the action is on weak ground.

The same test applies to procedural changes. Rewriting a procedure may be appropriate if the investigation establishes that the procedure was inaccurate, unavailable, poorly designed or inconsistent with the task. Rewriting a perfectly adequate procedure because someone did not follow it is a different proposition. Unless the investigation understands why the deviation occurred, the organisation may simply produce another version of a document that people will eventually work around for the same reasons.

The control question changes the investigation

A useful way to improve the connection between investigation and corrective action is to make controls central to the analysis.

Instead of moving directly from “what did the person do?” to “how do we stop them doing it again?”, investigators can ask:

• What should have prevented this event?

• Was that defence present?

• Did it operate as intended?

• If it failed, how did it fail?

• If it was absent, why was it absent?

• What organisational conditions affected its design, implementation, maintenance or verification?

These questions shift attention from correcting behaviour to understanding how risk was supposed to be controlled.

That is important because sustainable corrective actions usually need to strengthen the system's ability to manage the hazard. We require corrective actions to address the relevant Absent or Failed Defences and Organisational Factors, reduce the likelihood of recurrence and improve system defences. It also directs investigators to consider the Hierarchy of Controls rather than relying automatically on lower-order administrative responses.

The practical difference can be significant.

Suppose an operator enters an equipment interaction zone because they believe a machine has been isolated. An investigation focused primarily on the operator may recommend refresher isolation training and a reminder about exclusion zones.

A control-focused investigation may establish that the isolation status was difficult to verify from the work position, the design allowed stored energy to remain, the permit process did not require independent verification for that task, and previous reports about confusing isolation points had not resulted in engineering review.

Now the corrective action discussion has substance.

Training might still form part of the response, but it is no longer carrying the entire burden of prevention.

Corrective action quality is not the same as action-register quality

There is another trap worth addressing.

Organisations can become very good at managing actions without becoming good at designing them.

An action register can tell you whether an action has an owner. It can tell you when the action is due, whether it is overdue and whether someone has uploaded evidence of completion.

Those are useful management functions. They do not answer the most important question:

Will this action materially reduce the risk identified by the investigation?

We use the SMARTER approach for corrective actions: Specific, Measurable, Accountable, Reasonable, Timely, Effective and Reviewed.

The last two elements deserve particular attention.

An action is not effective simply because it has been completed. Completion tells you that the planned activity occurred. Effectiveness asks whether the action produced the intended improvement in the control environment and reduced the relevant risk.

Review then asks whether that improvement has been sustained.

An organisation can achieve a 100 per cent action closure rate while repeatedly closing actions that do little to strengthen its defences. From a dashboard perspective, performance looks excellent. From a risk perspective, very little may have changed.

Sustainability starts with diagnosis

When organisations talk about sustainable corrective actions, the discussion often moves immediately to implementation: leadership support, accountability, resources, communication and follow-up.

All of those can matter. But sustainability starts earlier.

A sustainable action needs to be connected to a correctly identified contributing factor. That contributing factor needs to be supported by evidence. The investigation needs to understand how the factor affected the event and what control weakness allowed the risk to develop.

Only then can the organisation make a sensible decision about what should change.

This is why improving corrective actions without improving investigation quality has limited value. Investigators can be taught to write better SMARTER statements, assign clearer owners and develop stronger effectiveness measures. Those are useful skills, but they cannot compensate for an investigation that has misunderstood the event.

Better wording does not repair weak analysis.

A practical quality test

Before approving a corrective action, work backwards.

Start with the proposed action and ask what finding it addresses. Then identify the contributing factor behind that finding. Locate the evidence supporting the contributing factor. Confirm how the analysis established its relevance to the incident. Finally, ask what defence the action strengthens and how the organisation will know whether the intended risk reduction has occurred.

If that chain breaks at any point, the corrective action deserves another look.

This is particularly useful when reviewing familiar actions such as “provide refresher training”, “update the procedure”, “remind employees”, “increase supervision” or “conduct a toolbox talk”.

For example, if the recommendation is additional training, ask what evidence established a competency gap. If the recommendation is a revised procedure, ask what evidence established that the existing procedure contributed to the event. If the recommendation is increased supervision, ask what aspect of supervision failed and what organisational conditions influenced that failure.

Those questions are not intended to make recommendation development difficult. They protect the organisation from spending time and money implementing actions that cannot be traced back to the investigation.

Strong corrective actions are traceable

A high-quality investigation creates a visible line between evidence and action.

The evidence establishes what occurred. PEEPO helps organise and test the available information. Timeline reconstruction helps the investigation understand the sequence and identify missing or conflicting information. Contributing factors are transferred into the ICAM analysis and considered as Absent or Failed Defences, Individual or Team Actions, Task or Environmental Conditions and Organisational Factors.

Findings should emerge from that analysis rather than being added afterwards because they sound reasonable.

Corrective actions then address those findings.

That traceability is one of the strongest quality controls available to an investigation team. It makes assumptions easier to challenge, exposes recommendations that have no evidentiary basis and gives managers a clearer reason for investing in the proposed change.

It also improves learning.

If several investigations identify similar Organisational Factors, the organisation can begin to see patterns beyond individual events. What appeared to be separate incidents may reveal common weaknesses in change management, equipment design, procedural systems, contractor interfaces or operational planning.

That is where investigation moves beyond incident closure and starts contributing to organisational learning.

What leaders should look for before approving the report

Leaders do not need to rerun the entire investigation to test whether corrective actions are credible. A few focused questions can reveal a great deal about the quality of the analysis.

When an action recommends training, ask what evidence demonstrates that knowledge or competency contributed to the event. When a procedure is being rewritten, ask what was wrong with the existing procedure and how that affected the work. When supervision is being increased, ask what failed in the existing supervision system and whether workload, role clarity or verification requirements contributed.

For engineering actions, ask which Absent or Failed Defence the proposed change strengthens. For every significant action, ask how effectiveness will be measured after implementation rather than relying only on evidence that the action has been completed.

Most importantly, ask whether the investigation has identified the organisational conditions that influenced the event.

The purpose is not to force every investigation to produce a long list of Organisational Factors. More Organisational Factors do not automatically mean a better investigation. They require sufficient evidence and should describe specific ineffective risk management processes that contributed to the event.

The objective is quality of analysis, not quantity of findings.

Stop judging corrective actions only at the end

When a corrective action fails, it is reasonable to examine implementation. Perhaps the action was poorly executed, insufficiently resourced or never properly embedded.

But that review should also travel upstream.

Was the right evidence gathered? Were assumptions tested? Did the timeline expose gaps or contradictions? Were the relevant Absent or Failed Defences identified? Did the analysis move beyond the actions of the people closest to the event? Were Organisational Factors supported by evidence and clearly connected to the incident? Did each corrective action address an established contributing factor?

These questions change how organisations think about corrective action failure.

Sometimes the fix failed because implementation was weak. Sometimes the fix failed because the organisation never understood what needed fixing.

That distinction matters. If the diagnosis is wrong, stronger project management will not save the corrective action. More reminders will not save it. A better action tracking system will not save it either.

The investigation has to give the organisation a reliable picture of how its controls were defeated and what conditions allowed that failure to develop. Only then can corrective actions be designed to strengthen the system in a way that has a realistic chance of lasting.

A closed action is an administrative result.

A stronger defence is the result that matters.