How to ruin an
accident investigation

A report is written, employees are interviewed, documents are reviewed and corrective actions are recorded. Yet if completing the report becomes the objective instead of understanding the event, the organisation may identify one visible error and continue operating almost as before.

I will examine four common ways to ruin an accident or near-miss investigation. The example is a real near miss in which I led the investigation.

The case: dangerous work seven metres above the floor

A factory used a large process trolley that travelled on rails approximately seven metres above the floor near the roof. One day the trolley stopped during production and required repair. There was no fixed access route. Employees solved the problem by leaning long ladders against the trolley and climbing up to perform the repair. Using ladders in this way was prohibited.

During the work, a tool made contact with a limit switch. The trolley interpreted the contact as a command to move, shifted and knocked the ladders over. The employee avoided falling only because the ladders became caught on a ventilation duct. The movement was possible because the trolley had not been properly isolated from its energy source. The lockout/tagout procedure had not been completed.

The consequences could have been extremely serious.

The first question that comes to mind is easy:

Why did the employee use a prohibited ladder?

The question matters, but by itself it is not enough. If the investigation ends there, the visible act is mistaken for the whole explanation.

01Investigation mistake 1: the scene is treated as a formality

The first way to ruin an investigation is to visit the scene merely because the procedure requires it.

The physical environment is not background scenery. It is evidence. At the scene, investigators should establish the actual positions, distances, routes, controls, tools, lighting, visibility and physical restrictions. They should reconstruct what the employee could see, reach and do at each point in the event.

In this case, the scene revealed that the component was high above the floor and had no permanent access. It also showed why the available scissor lift was not considered practical for this specific task. The equipment, layout and production environment shaped the options available to employees.

A photograph taken from the floor does not provide the same understanding as examining the work from the employee’s position. If safe access is difficult, slow or unreliable, that fact is not a minor detail. It may be one of the conditions that created the unsafe practice.

A scene visit should test the emerging explanation, not merely illustrate a report that has already been decided.

02Investigation mistake 2: the interview becomes an interrogation

The second way to ruin an investigation is to enter the interview with a conclusion already in mind.

If the investigator’s tone communicates blame, employees begin to protect themselves. They shorten answers, avoid uncertainty and tell the version they believe the organisation wants to hear. The result may appear precise while being less truthful and less useful.

Questions such as “Why did you break the rule?” assume both motive and guilt. A better starting point is to ask the employee to describe the work from the beginning: what the situation was, what information was available, what options were considered and what made the chosen method seem workable at the time.

In this case, the employee explained that using the lift was regarded as too slow. The fault needed to be removed quickly so production could continue. The ladders were nearby and provided immediate access.

This did not make the method acceptable. It did explain how the method emerged and why it persisted. The purpose of an interview is not to remove responsibility. It is to collect reliable information before conclusions and actions are selected.

03Investigation mistake 3: documents are confused with real work

The third way to ruin an investigation is to assume that instructions, risk assessments and training records describe actual work.

Everything may look correct on paper. Dangerous ladder use is prohibited, employees have been trained, a personnel lift is available and procedures for safe isolation exist. The task risks may also have been formally assessed.

On that evidence, the near miss looks like one employee choosing to ignore the rules. But documents describe how work is supposed to happen. They do not prove that work can or does happen that way.

If an instruction requires a lift, the investigation must also establish whether the lift is genuinely available, whether it can reach the location, and how much its deployment delays the task. When the safe method is continually difficult or practically impossible, employees begin to develop their own solutions.

The organisation may then have designed safe work on paper, but not in practice.

The employee suggested that a screwdriver used in the repair might have touched the limit switch and initiated movement. An interview statement alone did not prove the sequence. It had to be compared with physical observations, technical information and the trolley’s operation.

The movement nevertheless showed that hazardous energy had not been reliably isolated before the work began.

This is a central investigation principle. A single observation should not be ignored, but neither should it become an instant conclusion. Documents, physical evidence and employee accounts must be examined together. A conflict between them may be the most important finding in the entire investigation.

04Investigation mistake 4: analysis ends with human error

The fourth and often decisive mistake is to stop the analysis as soon as a clear deviation is identified.

The employee used ladders even though their use was prohibited. That may be entirely true, but it does not yet explain the event. If the analysis ends there, the actions are predictable: remind the employee, retrain the workforce, emphasise the prohibition and ask supervisors to monitor compliance.

These actions look credible in a report. They do not remove the conditions that produced the unsafe method.

The analysis must continue. Why was the safer equipment impractical? Why was there no safe access route? Why had a recurring fault not been given a permanent work method? Was the unsafe practice known to others? How did production needs influence the decision? Why was hazardous energy not isolated?

The employee’s action was the most visible part of the sequence. Behind it were equipment design, access to tools, production pressure, the lack of permanent access and shortcomings in hazardous-energy control. Only by identifying these factors could the organisation choose effective corrective actions.

What corrective actions prevent recurrence?

A good investigation does not end by asking people to be more careful. It changes the work system so that safe action becomes easier and unsafe action becomes harder or less attractive.

In the trolley case, dangerous climbing was stopped and the long ladders were removed from the area. This immediately made continuing the old practice more difficult. A boom lift was obtained because it provided better access than the scissor lift. Practical arrangements also made it faster to deploy, so that the safe method would not keep losing to the quicker shortcut.

The method also had to include reliable isolation of hazardous energy before troubleshooting began. Safe physical access is not enough if the machine or connected system can move or start during the work.

In the longer term, fixed maintenance platforms were built and difficult-to-reach components were relocated lower down. Inspection and maintenance could then be performed safely. The learning was not limited to one trolley or one factory: maintenance practices were developed across all company sites.

Four questions for a good investigation

A successful accident or near-miss investigation combines observations from the scene, interviews, documents and systematic analysis.

It should be able to answer four questions:

1. What happened?
2. Why did the action appear reasonable or possible in that situation?
3. Which factors in the work system enabled the event?
4. What must change so that safe work succeeds in practice?

The quality of corrective actions can be tested with one further question:

Does the action remove a condition that enabled the event, or does it merely ask a person to perform better in the same conditions?

Training, instructions and supervision all have a place. They are weak solutions, however, when tools, equipment design or work organisation continually push people in another direction.

The purpose is not a tidy explanation

Accidents and near misses are rarely caused by one poor decision alone. They usually involve several technical, organisational and human factors.

If an employee error is found quickly, the report can be completed efficiently. Responsibility looks clear and corrective actions are easy to record.

The organisation may nevertheless learn nothing.

Real learning often starts only after the obvious explanation has been found. Investigators must then determine why the action was possible, how it became normal practice and why the work system did not offer a better way to complete the task.