Summary
In short
- The skill is observation. Assessors who watch the task, ask what happens when it goes wrong, and notice the improvised tool produce useful assessments regardless of which matrix they use.
- Scoring consistency between assessors is worth calibrating and is not worth teaching at length. Worked examples do more than matrix theory.
- Training should cover why assessments fail: written from the procedure, non-routine modes omitted, frequency ignored, controls defaulting to PPE, residual risk unaccepted.
- Assess by having the trainee produce an assessment of a real task, not by a multiple choice test on the hierarchy of controls.
- ISO 45001 requires competence to be achieved and the effectiveness of the action evaluated, which a course completion certificate does not do.
- Where an organisation's assessments are consistently thin, the cause is usually training that taught paperwork rather than observation.
What it is
What it is
What does risk assessment training cover?
How to identify hazards by observing work as performed, how to evaluate risk consistently, how to apply the hierarchy of controls, and how to write an assessment that someone else can use. The scoring method is a small part, and the transferable skill is observation.
How should it be assessed?
By having the trainee produce a real assessment of a real task, reviewed against criteria. A knowledge check on the matrix and the hierarchy confirms recall. Producing an assessment that another person could work from confirms the skill the training exists to build.
When to use it
When to use it, and when not to
This records training in how to perform risk assessments. It is not the assessments produced.
Use it for
- Training people who will write or review risk assessments
- Refresher and calibration where assessments vary widely between assessors
- After a change of methodology, matrix or documentation system
- Where audit or incident findings indicate assessments are not identifying real exposures
- Training specific method skills: manual handling assessment, machinery assessment, quantified methods
Not for
- The risk assessments themselves, which are the output rather than the record of training
- Task-specific competence, which is assessed separately by observation of the task
- Awareness training for people who will follow assessments rather than write them
- The competency framework, which defines who requires this training
- Auditor training, which covers a related but distinct set of skills
Standards
What it is built against
Risk assessment competence sits under general competence requirements rather than a dedicated standard.
| Clause | Requirement | Where it lands |
|---|---|---|
| ISO 45001 cl.7.2 | Competence determined and achieved, with the effectiveness of actions taken evaluated | Outcome |
| ISO 45001 cl.6.1.2 | Hazard identification proactive and covering routine and non-routine activities and human factors | Theory content |
| ISO 45001 cl.8.1.2 | Hierarchy of controls, which trainees must be able to apply rather than recite | Methods covered |
| ISO 31000 cl.6 | Risk assessment process within an established context, applied consistently | Theory content |
| HSE Management Regs reg.7 | Competent persons appointed to assist with health and safety measures | Header |
| ISO 45001 cl.5.4 | Worker participation in hazard identification, which assessors must know how to enable | Practical exercise |
| ISO 11228 series | Specific methods for manual handling assessment, where those apply | Methods covered |
| ISO 12100 | Machinery risk assessment methodology, a distinct skill requiring separate coverage | Methods covered |
What it does not cover
- The assessments produced, which are the output and are recorded separately.
- Task-specific competence, assessed by observing the person perform the task rather than assess it.
- Awareness training for people who follow assessments rather than write them.
- The competency framework, which determines who needs this training.
- Auditor training, which covers overlapping but distinct skills.
Filling it in
Filling it in well
Teach observation, calibrate scoring rather than lecturing it, and assess with a real assessment.
How to watch a task, what to look for, how to ask what happens when it goes wrong, how to notice improvised tools and worn access points, and why the night shift version differs. This is the part that transfers and the part most courses treat as a preamble to the matrix.
Give the group the same scenario and compare ratings. The discussion about why one person scored it four and another three teaches more about consistency than any amount of matrix theory, and it produces the alignment that makes assessments comparable across a site.
Written from the procedure, non-routine modes omitted, frequency ignored, controls defaulting to training and PPE, residual risk unaccepted, and never reviewed after change. Naming these gives trainees a checklist against their own work and is more useful than any amount of methodology.
Have the trainee assess an actual task and review the output against criteria: did they observe it, did they cover non-routine modes, did they apply the hierarchy honestly, could someone else work from what they wrote. That evaluates effectiveness, which the standard requires and a certificate does not.
Audit findings
Common audit findings
Findings here surface as thin assessments rather than as training records.
| Finding | Clause | What fixes it |
|---|---|---|
| Training covers the matrix and paperwork with little on observation. | ISO 45001 cl.6.1.2 | Weight the content toward watching work; that is the transferable skill. |
| Assessed by knowledge test rather than by producing an assessment. | ISO 45001 cl.7.2 | Have trainees assess a real task; effectiveness must be evaluated. |
| No calibration between assessors, producing wide scoring variation. | ISO 31000 cl.6 | Run shared scenarios and compare; discussion produces alignment faster than theory. |
| Method-specific skills not covered where those methods are required. | ISO 11228 / ISO 12100 | Manual handling and machinery assessment are distinct methodologies needing their own training. |
| Trainees not taught how to involve the people doing the work. | ISO 45001 cl.5.4 | Participation is a requirement and a technique; teach how to ask. |
| Common failure modes not covered. | ISO 45001 cl.6.1.2 | Name them; a list of how assessments go wrong is a practical self-check. |
| No refresher despite a change of methodology or matrix. | ISO 45001 cl.7.2 | Methodology change is a retraining trigger; old habits persist against new forms. |
| Thin assessments across the site with no review of the training. | ISO 45001 cl.9.1 | Uniformly weak output is a training finding, not an individual one. |
| Reviewers of assessments not trained to the same level as writers. | ISO 45001 cl.7.2 | A reviewer who cannot spot the failure modes approves them. |
| Training delivered once with no ongoing calibration. | ISO 31000 cl.6 | Consistency drifts; periodic calibration maintains comparability. |
Worked case
Case in point: everyone scored it differently
A site trained thirty supervisors in risk assessment over three sessions. The course covered the hierarchy of controls, the five-by-five matrix and the site's documentation. Knowledge checks were passed and certificates issued.
A year later a review of assessments found wide variation. The same task assessed by different supervisors carried ratings from four to sixteen, controls ranged from full guarding to a toolbox talk, and roughly half the assessments had been written without anyone watching the task.
A half-day calibration session was run instead of a further course. Groups assessed the same two tasks, compared results, and discussed the differences. Variation on subsequent assessments narrowed considerably, and the proportion written from observation rose because the discussion made the difference visible.
Definitions
Definitions and key terms
- Calibration
- Aligning assessors by having them assess the same scenario and compare, producing comparable ratings across a site.
- Observation skill
- The ability to watch work as performed and notice deviations, workarounds and improvised tools.
- Failure modes
- The recurring ways assessments go wrong, useful as a self-check for anyone writing one.
- Effectiveness evaluation
- The ISO 45001 requirement to evaluate whether training achieved competence, distinct from recording attendance.
- Quantified method
- A validated tool such as the NIOSH equation or MAC producing a numerical exposure figure rather than a subjective rating.
- Risk matrix
- A grid combining likelihood and severity, useful for prioritisation and less important than the observation behind it.
- Reviewer competence
- The capability of those approving assessments, which needs to be at least equal to that of the writers.
- Assessment criteria
- What a trained assessor's output must demonstrate, used to evaluate the practical exercise.
FAQ
Frequently asked questions
What should risk assessment training concentrate on?+
Observation. How to watch a task, how to ask what happens when it goes wrong, how to notice improvised tools and worn access points, and why the version performed on nights differs. Two competent assessors will disagree about a rating and agree about the hazard and the control, which tells you where the value sits.
How should trainees be assessed?+
By producing an assessment of a real task, reviewed against criteria: did they observe it, did they cover the non-routine modes, did they apply the hierarchy honestly, and could another person work from what they wrote. ISO 45001 requires the effectiveness of the action to be evaluated, and a knowledge test on the hierarchy does not evaluate the skill.
What do we do about inconsistent scoring?+
Calibrate rather than retrain. Give a group the same task, have them assess it independently, then compare and discuss the differences. The argument about why one person scored four and another sixteen produces alignment quickly, costs half a day, and can be repeated as consistency drifts.
Should reviewers be trained too?+
To at least the same level. A reviewer who cannot recognise an assessment written from the procedure, or one that omits the non-routine modes, will approve it. In many organisations the approval step is performed by someone with less training than the writer, which converts review into a signature.
What if our assessments are uniformly weak?+
Treat it as a training finding rather than an individual one. Where the whole population produces thin assessments, the common factor is what they were taught, and the usual cause is a course that covered the form and the matrix rather than how to watch work and ask questions about it.
The agents
What the agents do with it
The training produces assessors. What fails is a course about the form, and reviewers who cannot see what is missing.
Records the practical exercise and its review against criteria, rather than a knowledge check, so effectiveness is actually evaluated.
Compares assessment quality and scoring across assessors, surfacing calibration drift before it shows up in an audit.
Connects thin or aged assessments back to who wrote them, so training gaps appear as a pattern rather than as individual findings.
Covers quantified method training where manual handling or posture assessment requires a validated tool rather than a matrix.
This template lives in KnowTrain — training and credentials. Induction, competency, toolbox talks, refreshers and expiry tracking.
Meet KnowTrain→Sources
Sources
- ISO 45001:2018 clauses 6.1.2, 7.2 and 8.1.2
- ISO 31000:2018, risk management guidelines
- Management of Health and Safety at Work Regulations 1999, regulations 3 and 7 (GB)
- ISO 11228 series, ergonomics, manual handling
- ISO 12100:2010, safety of machinery, risk assessment and risk reduction