How to investigate a workplace accident: finding the root cause, not the blameHow to investigate a workplace accident: finding the root cause, not the blame

When something goes wrong at work, there is a strong pull toward the quickest explanation: the worker was careless, they rushed, they didn’t follow the rules. It feels like an answer, and it closes the file. It is also almost never the real cause, and an investigation that stops there guarantees the same thing happens again to someone else. This guide sets out how to investigate a workplace accident properly, using the HSE’s own framework.

The prize is not paperwork. It is the third word of Record. Resolve. Prevent. That means turning an incident into the specific, lasting change that stops the next one.

Why investigate at all?

There are three reasons, and they reinforce each other:

• To prevent recurrence: the only reason that actually matters. Every accident is a lesson already paid for; the investigation is how you collect it.
• To meet your legal duties: investigating helps you show you are managing risk, and it underpins accurate RIDDOR reporting where the incident is reportable.
• To protect the organisation: a thorough, contemporaneous investigation is your evidence if a claim or prosecution follows months later.

The four steps of an investigation

The HSE’s guidance HSG245, “Investigating accidents and incidents”, sets out a four-step structure that scales from a near miss to a serious injury:

1. Gather the information. Collect the facts while they are fresh: the scene, photographs, positions, equipment, and accounts from those involved and any witnesses. Physical evidence and memories both decay fast, so this step is urgent.
2. Analyse the information. Work out what happened and why, separating the immediate cause from the deeper ones (see below). This is where most investigations either succeed or give up too early.
3. Identify risk control measures. Decide what needs to change so it cannot happen again, choosing controls high up the hierarchy rather than another sign or another briefing.
4. Action and implement. Turn those measures into a plan with owners and deadlines, then check they were actually done and that they worked.

The three levels of cause

The heart of a good investigation is refusing to stop at the obvious. HSG245 asks you to find three layers of cause:

• Immediate cause: the most obvious agent of harm. The guard was missing; the worker slipped; the load fell.
• Underlying cause: the less obvious condition that let the immediate cause exist. The guard was removed because it slowed the job and nobody replaced it; the floor was wet because a leak went unreported.
• Root cause: the failing in the management system from which the rest springs. There was no system to check guards were refitted; there was no easy way to report the leak, so nobody did.

The immediate cause tells you what happened. The root cause tells you why your organisation let it happen, and it is the only level at which a fix actually prevents a repeat.

The 5 Whys: a simple tool that works

You do not need complex methodology to reach a root cause. The 5 Whys (asking “why?” of each answer until you run out of system failures to expose) is often enough:

1. A worker cut their hand on a machine. Why? The guard wasn’t in place.
2. Why? It had been removed to clear a jam and not refitted.
3. Why? Clearing jams by hand is quicker, and it’s done several times a shift.
4. Why? The machine jams often and nobody has reported it as a problem.
5. Why? Reporting a recurring fault is a hassle, so people just work around it.

“The worker was careless” would have stopped at the first line. Five questions later the real fix is obvious: make the machine jam less, and make reporting the fault easier than working around it. That is a root-cause fix; “be more careful” is not.

Investigate the system, not the person

The fastest way to kill your own investigations is to use them to assign blame. The moment workers believe that reporting an incident, or telling the truth about one, will get someone disciplined, the information dries up, and information is the entire raw material of prevention. A just culture distinguishes an honest mistake from genuine recklessness, and treats the first as a chance to learn. It is not softness; it is the only way to keep the near misses and reports flowing that let you see risk before it becomes injury.

The investigation is only as good as the data

Step one (gather the information) is where most investigations are quietly lost. A report scribbled in a notebook hours later, with no photo of the scene, no exact time, no location, and a fading memory of who was where, cannot support a serious analysis. For a field workforce it is harder still: the incident happens miles from the office, and by the time anyone with a clipboard arrives, the scene is gone.

This is precisely the gap jobsafe closes. The person who was there captures the incident in seconds, on the phone in their pocket (photos of the scene, automatic location and timestamp, an account while it is fresh), and it reaches a supervisor instantly. From there the built-in resolution flow carries the report from raised, to investigated, to closed, with actions assigned and a full audit trail, so the “action and implement” step doesn’t quietly evaporate. Watch it in the academy, and read what RIDDOR requires when an investigated incident crosses the reporting threshold.

Sources and further reading

• HSE, HSG245: Investigating accidents and incidents (opens in new tab)
• HSE, Investigating accidents and incidents (overview) (opens in new tab)
• HSE, Root causes and human factors (opens in new tab)
• HSE, RIDDOR: reporting requirements (opens in new tab)