The Risk Assessment That Failed Before the Incident: What Investigations Reveal About Broken RA Processes
What investigations reveal about broken risk assessment processes
The assessment was four pages, signed by seven people, identical to three other sites. The site-specific hazard that killed the worker wasn't in it.
What every HSE Director and COO needs to understand
Every serious incident investigation eventually arrives at the same finding: the risk assessment for the activity was either missing, incomplete, outdated, generic, or had been copied from a template without site-specific adaptation. The risk assessment did not fail during the incident — it failed before the incident, often months or years earlier. This article argues that the chronic inadequacy of risk assessments in high-hazard industries is not a training problem. It is a management system design problem. Organisations treat risk assessment as a compliance document rather than a decision-making tool — and the result is a library of paperwork that protects the filing system but
THE NUMBERS THAT MATTER
The platform that had a risk assessment no one had read
A maintenance worker fell through a temporary platform on an industrial site. The platform had been erected three weeks earlier to provide access for a shutdown activity. A risk assessment existed for the platform installation. It was four pages long, contained 23 hazards, and had been signed by seven people. It was also identical to the risk assessment used for platform installations at three other sites in the network, with only the site name changed. The site-specific hazard — a gap between the platform edge and the structure it was attached to — was not identified. The gap was visible to anyone standing on the platform. It was not in the risk assessment because the risk assessment had never been conducted on site. It had been completed in an office, using a template, by a person who had not visited the work location. The investigation found that the platform risk assessment had been "completed" as an administrative requirement before the work permit was issued. It had not been used as a tool to identify and control actual hazards at the actual work location. The risk assessment was compliant. The worker was dead.
THE COMPLIANCE TRAP When risk assessment becomes a precondition for work authorisation rather than a tool for hazard identification, it ceases to protect workers. It protects the management system from regulatory criticism — at the cost of the workers it was designed to protect.
Five failure modes of risk assessment
-
Template risk assessments without site adaptation: Generic risk assessments copied from templates without site-specific hazard identification. The document exists, the hazards do not match reality.
-
Risk assessments completed in offices, not at the work face: The risk assessment is produced by someone who has not physically inspected the work location, equipment, or conditions. The gap between the document and reality is the gap through which people fall.
-
Outdated risk assessments for changed conditions: The work environment changes — layout modifications, new equipment, different weather conditions — but the risk assessment is not updated. The document reflects a past reality that no longer exists.
-
Risk assessments signed but not understood: Workers sign the risk assessment as a precondition for starting work, without reading or understanding its contents. The signature is a compliance record, not evidence of comprehension or agreement.
-
Risk ratings manipulated to avoid controls: Hazards are rated as "low" or "medium" to avoid triggering additional control requirements. This is particularly common when production pressure makes additional controls inconvenient. The risk matrix becomes a tool for justifying inaction, not for driving protection.
KEY TAKEAWAY If your last 10 incident investigations identified "inadequate risk assessment" as a contributing factor, the problem is not individual competence — it is the management system's design of the risk assessment process. You are treating a systemic failure as a training gap.
The risk assessment quality framework for practitioners
-
Physical inspection before documentation: No risk assessment may be completed without the assessor physically inspecting the work location, equipment, and conditions. Desktop assessments are prohibited for anything classified as high-risk work.
-
Site-specific hazard identification: Every risk assessment must identify at least three hazards that are unique to this specific task, at this specific location, on this specific day. If the hazard list is identical to the template, the assessment has not been conducted.
-
Worker participation in assessment: The workers who will perform the task must participate in the risk assessment. They know hazards that no office-based assessor will identify. Their participation is not optional — it is a quality requirement.
-
Dynamic review at key milestones: Risk assessments must be reviewed and updated at defined milestones: start of each shift, after any change in conditions, and before any deviation from the planned work method.
-
Independent audit of risk assessment quality: A sample of risk assessments should be audited quarterly by someone other than the author, checking for site-specificity, currency, worker participation, and realistic risk ratings.
THE PRACTITIONER TEST Take your current risk assessment for the highest-risk activity on your site. Walk to the work face. Compare what the document says to what you see. If there is a gap — any gap — your risk assessment process is producing compliance documents, not safety tools.
"The risk assessment that failed before the incident is the most honest indicator of your management system's maturity. If your risk assessments are templates, your safety is theoretical."
— Bruno Hounkpati
“The risk assessment that failed before the incident is the most honest indicator of your management system's maturity. If your risk assessments are templates, your safety is theoretical.”
— Bruno Hounkpati
References
ILO (2024). Nearly 3 million people die of work-related accidents and diseases.
BLS (2026). Census of Fatal Occupational Injuries — 2024.
CPWR (2024). Construction fatalities by business size.
Lloyd's Register Foundation (2024). World Risk Poll 2024.
ISO 31000:2018. Risk Management — Guidelines.
Reason, J. (1997). Managing the Risks of Organizational Accidents.
This article is published by HSESKILLS Ltd for educational and informational purposes only. Composite scenarios illustrate common investigation patterns and do not refer to any specific organisation unless explicitly named.
