Stop Looking for Someone to Blame: What Good Incident Investigations Should Actually Find

An incident occurs.

The area is secured. Notifications begin. Management wants answers. An investigation team is appointed. Photographs are taken, witnesses are interviewed, documents are collected and a timeline begins to emerge.

Eventually, somebody asks the question everyone wants answered:

What caused it?

It sounds straightforward.

Unfortunately, serious incidents rarely have straightforward answers.

A worker entered the wrong area. A supervisor failed to verify a control. An operator made an incorrect decision. A contractor did not follow the procedure. Someone did not recognise the hazard.

These explanations are attractive because they provide a clear endpoint for the investigation.

They are also frequently where the investigation should begin.

If a competent employee performed an action that, after the event, appears obviously wrong, the interesting question is not simply why they made a mistake. It is why that action made sense to them at the time.

What information did they have? What were they trying to achieve? What conditions were present? What had they experienced previously? What assumptions were they making? What pressures influenced the decision? How was the work normally performed?

These questions take us beyond identifying the person closest to the event.

They help us understand the system that produced it.

Hindsight Changes Everything

Incident investigators possess something the people involved in the event did not have.

We know what happened.

That knowledge changes how we interpret everything that came before it.

A warning sign that seemed insignificant at 10:00 becomes obvious after the failure at 10:15. A decision that appeared reasonable during the work suddenly looks reckless once we know the outcome. An ambiguous instruction becomes an obvious error.

This is hindsight bias.

It is one of the greatest challenges in incident investigation.

After an event, the path towards failure can appear almost inevitable. We connect the warning signs, identify missed opportunities and wonder why nobody recognised what was happening.

Before the event, however, those same signals were mixed with hundreds of other pieces of information that did not result in failure.

The investigation therefore needs to reconstruct the world as it appeared before the outcome was known.

Instead of asking, “Why didn't they see the danger?”, ask:

What did they see, and what did they think it meant?

That subtle change in questioning can reveal considerably more.

Human Error Is Usually a Description, Not a Cause

“Human error” remains one of the most common conclusions in incident investigations.

Someone forgot something. Selected the wrong control. Failed to follow the procedure.

Misinterpreted information. Made a poor decision.

The problem is that identifying human error explains very little.

If an operator pressed the wrong button, we still need to understand why.

Were the controls confusing? Were two switches almost identical? Was the operator interrupted? Were they responding to several alarms simultaneously? Had the interface recently changed? Was the procedure unclear? Had the same action been performed successfully before?

Similarly, saying a worker “failed to follow procedure” does not explain why the procedure was not followed.

Was it known? Was it accessible? Was it practical? Did the task differ from what the procedure anticipated? Was deviation normal? Was the worker under pressure to complete the activity? Did supervisors routinely accept the same practice?

Human actions are part of the event.

They should not automatically become the end of the explanation.

The UK Health and Safety Executive's guidance on human factors makes this distinction particularly important. Human failures need to be understood in the context of the job, the individual and the organisation rather than being treated simply as personal shortcomings.

The stronger question is therefore not, “Who made the error?”

It is, “What made this error possible, likely or difficult to detect?”

Procedures Can Fail Too

One of the most predictable corrective actions after an incident is to revise the procedure.

Sometimes that is exactly the right response.

Sometimes it is almost meaningless.

A procedure can only reduce risk if people can understand it, access it and apply it within the conditions in which the work occurs.

If a worker deviated from a procedure, investigators should examine the procedure with the same scrutiny applied to the worker.

Was it clear?

Was it current?

Did it accurately describe the task?

Did it conflict with another requirement?

Could the work realistically be completed as written?

Were employees routinely adapting it?

If experienced workers have been performing the task differently for years and supervisors know about the practice, describing the incident as an isolated procedural violation misses something important.

The unofficial method may have become the real method.

The investigation needs to understand how that happened.

Look at Normal Work, Not Only the Incident

A powerful investigation technique is to compare the incident with occasions when the same activity was completed successfully.

How was the job normally performed?

Were the same shortcuts used?

Did the same equipment condition exist?

Was the same procedure routinely adapted?

Were supervisors normally aware?

If the behaviour involved in the incident occurred regularly without negative consequences, the explanation becomes more complex than one employee making a bad decision.

The organisation may have gradually accepted a way of working that appeared successful.

This is important because incidents often emerge from normal operations rather than from completely abnormal behaviour.

People do not generally arrive at work intending to create an accident. They use practices that have previously allowed them to achieve their objectives.

Most of those practices work.

Until one day the circumstances are slightly different.

Understanding normal work helps investigators identify what changed on the day of the event and, just as importantly, what had been present for months or years without attracting attention.

Ask What Conditions Shaped the Decision

Behaviour does not occur in isolation.

It is shaped by the environment around it.

Time pressure matters. Staffing matters. Fatigue matters. Equipment condition matters.

Supervision matters. Experience matters. Workload matters. Incentives matter. Design matters.

Consider a supervisor who authorised work despite an incomplete control.

It is easy to conclude that the supervisor demonstrated poor judgement.

But what else was happening?

Was the project significantly behind schedule? Had similar work proceeded previously under the same conditions? Was the supervisor responsible for several simultaneous activities? Was specialist support unavailable? Had management communicated strong expectations about completing the work that day?

None of these factors automatically excuses the decision.

That is an important distinction.

Understanding behaviour is not the same as removing accountability.

A mature investigation can examine individual responsibility while also examining the conditions that influenced behaviour.

If we stop at the individual, we may discipline one person while leaving the next person exposed to exactly the same pressures.

Root Cause Is Rarely a Single Thing

The language of “root cause” can unintentionally encourage organisations to search for one fundamental explanation.

Complex events rarely work that way.

A serious incident may involve equipment condition, planning weaknesses, unclear responsibilities, inadequate supervision, conflicting priorities, design limitations, communication failures and individual decisions.

Remove any one of these factors and the event might not have occurred.

Which one is the root?

Often, that is the wrong question.

The objective should be to identify the combination of conditions that allowed the event to develop and determine where intervention can most effectively reduce recurrence.

Techniques such as the Five Whys can be useful for relatively straightforward problems, particularly when they encourage investigators to move beyond the immediate failure.

But no investigation method should force a complex event into an artificially simple chain.

The quality of an investigation is not determined by whether it successfully identifies a box labelled “root cause.”

It is determined by whether the organisation understands the event well enough to make recurrence less likely.

Beware of Convenient Causes

Some investigation findings appear repeatedly because they are easy to identify and easy to close.

Lack of awareness.

Inadequate communication.

Failure to follow procedure.

Poor supervision.

Insufficient training.

These may be genuine contributors.

But they are often descriptions requiring further investigation.

If communication was inadequate, why?

If supervision was poor, what prevented effective supervision?

If somebody lacked awareness, what information should have created that awareness?

If training was inadequate, what capability was actually missing?

Convenient causes often lead to convenient corrective actions.

Conduct refresher training. Reissue the procedure. Deliver a toolbox talk. Send a safety alert.

The action tracker becomes green.

But the underlying conditions may remain untouched.

Corrective Actions Should Change the System

The quality of an investigation can often be judged by the quality of its corrective actions.

If a serious incident results primarily in additional training, reminders and procedural revisions, it is worth asking whether the investigation went deep enough.

Strong corrective actions address the conditions that made the event possible.

Can the hazard be eliminated?

Can the equipment be redesigned?

Can an engineered safeguard prevent the error?

Can the work sequence be changed?

Can the interface be simplified?

Can the system make the safe action easier and the unsafe action more difficult?

This is where the hierarchy of controls should influence incident investigation.

Administrative controls have an important role, but they depend heavily on continued human performance.

If an error can reasonably be engineered out of the system, telling people to “take more care” is a weak substitute.

Training Is Not the Universal Corrective Action

Training deserves particular attention because it appears so frequently in investigation reports.

Training is appropriate when the investigation identifies a genuine gap in knowledge or skill.

But many incidents do not occur because people did not know what they were supposed to do.

Sometimes they knew perfectly well.

They were unable to do it because equipment was unavailable, workload was excessive, the procedure did not reflect reality or operational pressure encouraged another course of action.

Sending somebody back to training under those circumstances solves nothing.

Before recommending training, investigators should be able to answer a simple question:

What specific knowledge or skill was missing, and how did that deficiency contribute to the event?

If there is no convincing answer, training probably is not the corrective action.

Learning Requires Psychological Safety

The quality of an investigation depends heavily on what people are willing to tell investigators.

If employees believe the investigation is primarily searching for somebody to punish, their behaviour is predictable.

They become cautious.

Language becomes formal. Memory becomes selective. Informal practices disappear from the story. People describe how work is supposed to happen rather than how it normally happens.

The investigation may still collect large amounts of information.

It just may not collect the information that matters most.

This is where a just culture becomes important.

A just culture does not mean nobody is accountable.

Reckless behaviour, deliberate violations and intentional misconduct may require consequences.

But honest mistakes, normal human variability and reasonable decisions made within flawed systems require a different response.

Employees need confidence that telling the truth about how work actually occurs will be used primarily to understand and improve the system.

Without that confidence, organisations can conduct extremely professional investigations into carefully sanitised versions of reality.

Leadership Behaviour After an Incident Matters

Employees pay close attention to how leaders react after something goes wrong.

If the first executive question is, “Who did this?”, the organisation learns something.

If the first question is, “How could our systems have allowed this to happen?”, it learns something different.

Neither question eliminates accountability.

But the order matters.

Senior leaders should resist the natural pressure for immediate conclusions.

Serious events create emotion, operational disruption, reputational concerns and sometimes regulatory pressure. There is understandable demand for answers.

Premature certainty can damage the investigation.

Leaders should create space for investigators to understand complexity while ensuring immediate risks are controlled.

They should also avoid communicating conclusions before the evidence supports them.

Once leadership publicly describes an event as “operator error” or “failure to follow procedure,” it becomes considerably harder for an investigation team to reach a different conclusion.

Near Misses May Be More Valuable Than Injuries

Organisations naturally investigate events according to actual consequence.

A serious injury receives substantial attention.

A near miss may receive much less.

That can be a mistake.

Actual consequence contains an element of circumstance.

A dropped object misses somebody by 30 centimetres. A vehicle stops one metre before striking a pedestrian. A gas release does not ignite. A worker falls but the fall-arrest system works.

The absence of injury does not mean the control failure was insignificant.

Potential consequence matters.

High-potential events can provide organisations with an opportunity to learn from the conditions capable of producing serious harm without paying the human price of the worst outcome.

They deserve investigation proportional to what could reasonably have happened, not simply what did happen.

Investigations Should Look Upwards as Well as Downwards

Incident investigations naturally concentrate on the workplace where the event occurred.

That is necessary.

But some of the most influential decisions may have been made much higher in the organisation.

Who established the staffing level?

Who approved the schedule?

Who selected the contractor?

Who decided to defer maintenance?

Who designed the equipment?

Who determined the training requirements?

Who decided the risk was acceptable?

The purpose is not to find a more senior person to blame.

It is to understand how organisational decisions created the conditions experienced at the frontline.

Risk flows through organisations.

So should investigations.

Close the Learning Loop

Producing the investigation report is not the end of the process.

Neither is closing the corrective actions.

The organisation needs to establish whether the actions actually worked.

Did the engineering modification remove the exposure?

Did the redesigned procedure improve the task?

Did supervision become more effective?

Has the same finding appeared elsewhere?

Are employees still using the workaround that contributed to the event?

Without follow-up, organisations can repeatedly investigate variations of the same problem while believing previous actions were successfully completed.

Learning therefore requires verification.

The question is not simply whether the action was closed.

It is whether the risk changed.

Share Lessons, Not Just Safety Alerts

After significant incidents, organisations frequently issue safety alerts.

A photograph appears at the top. The event is summarised. Immediate causes are listed.

Several lessons are provided.

The alert is emailed across the business.

That is communication.

It is not necessarily organisational learning.

Learning requires people to determine whether the conditions that produced the event exist elsewhere.

Could this happen at our site?

Do we use similar equipment?

Do our procedures contain the same weakness?

Do our contractors face similar pressures?

Do we rely on the same control?

This turns an incident from somebody else's story into a prompt for local examination.

A good lesson learned should cause another part of the organisation to look at itself.

The Objective Is Not to Explain the Past

Incident investigation naturally looks backwards.

But its purpose is forward-looking.

We cannot change what happened.

We can change the conditions under which future work occurs.

That requires moving beyond simple explanations.

“Human error” is not enough.

“Failure to follow procedure” is not enough.

“Lack of awareness” is not enough.

Good investigations reconstruct the circumstances surrounding the event. They examine normal work. They understand why decisions made sense. They identify technical, organisational and human contributors. They develop corrective actions that change the conditions capable of producing recurrence.

Most importantly, they create learning without pretending accountability and learning are opposites.

When somebody makes a mistake, we should absolutely understand what they did.

Then we should keep investigating.

Because the person closest to the incident is often the easiest part of the system to see.

The conditions that placed them there are usually much more important.

And if those conditions remain unchanged, replacing, retraining or disciplining the individual may simply ensure that somebody else eventually makes the same mistake.

References and Further Reading

UK Health and Safety Executive (HSE). Investigating Accidents and Incidents: A Workbook for Employers, Unions, Safety Representatives and Safety Professionals (HSG245). Practical guidance on investigating adverse events, identifying immediate, underlying and root causes, and developing appropriate risk-control measures.

UK Health and Safety Executive (HSE). Reducing Error and Influencing Behaviour (HSG48). Guidance on human factors, human failure and the job, individual and organisational factors that influence behaviour and performance.

US Occupational Safety and Health Administration (OSHA). Incident Investigation. Guidance emphasising identification of underlying hazards and system weaknesses rather than investigation approaches focused primarily on assigning blame.

International Association of Oil & Gas Producers (IOGP). Learning From Normal Work. Human-performance resources encouraging organisations to understand how work is actually performed and to learn from both successful operations and unwanted events.

Energy Institute. Human and Organisational Factors resources. Industry guidance addressing human performance, organisational conditions, learning, investigation and systemic contributors to major accident risk.

Dekker, S. The Field Guide to Understanding 'Human Error'. CRC Press. Examination of hindsight bias, local rationality and systemic approaches to understanding human contribution to unwanted events.