"Suitable and sufficient" is the statutory test in regulation 3 of the Management of Health and Safety at Work Regulations 1999, and in practice most arguments about it come down to one question: is this assessment about this site, or about this activity in general?
A generic assessment can be perfectly well written and still fail, because it assesses a category of work rather than the work being done. Here is what closes that gap.
What makes an assessment genuinely site-specific
The hazards come from the actual scope
Not from a library of hazards for your trade. Take the tasks you are actually doing, in the sequence you are doing them, and ask what could cause harm at each stage. The test is directional: every hazard should be traceable to something in the scope, and every activity in the scope should produce at least one hazard.
The commonest failure is the reverse — a hazard register that arrived complete and was pruned. That produces documents listing confined space entry on a job with no confined space, which tells a reader more about the template than the site.
The controls are things that exist here
"Use mechanical lifting aids" is a category. "Use the client's pallet truck, available from the goods lift lobby, for moving units from the loading bay to the riser" is a control. The second can be checked, briefed and enforced; the first cannot, and on a site where no pallet truck exists it is simply untrue.
Site-specific controls also account for constraints the site imposes: restricted delivery hours, a shared scaffold, a corridor that is an escape route, an occupier's rules about noise before nine.
The people are named
Supervisor, first aider, the competent person for any statutory inspection, the appointed person for lifting. With cards and expiry dates where the work requires them. Roles without names are the single fastest way to tell an assessment was not written for a project, because a project has people on it.
The emergency arrangements were verified, not assumed
Somebody walked to the muster point. Somebody checked which A&E is nearest and wrote its address down. Somebody confirmed the first aider is on this shift. Where the work is at height, in a confined space or in an excavation, somebody wrote a rescue plan that names a trained person and equipment that is on site.
The site information you were given is visible in it
If the pre-construction information contained an asbestos survey, a service drawing or a ground investigation, the assessment should reference it by title and date. An assessment that could have been written without reading the pack usually was. Pre-construction information covers what should be in the pack and what to do when it never arrives.
The five signals that read as generic
- Roles with no names — "site supervisor", "site first aider", "competent person".
- An emergency section that does not match the site: a muster point that is not here, a hospital named for a different town, or the phrase "as per site rules" where an arrangement should be.
- Hazards that do not correspond to the scope, in either direction — extras that do not apply, or activities that generate none.
- A residual risk column that is blank, or identical the whole way down.
- Traces of the last project: another site's name, an old client, a plant list that does not match, a revision history describing changes to a different job.
Two of those together and a reviewer stops reading. They are also, usefully, all findable by search before you submit — open the file and search for the previous project's name. It is there more often than anyone would like.
Cross-checking the assessment against the method statement
The most valuable review step, and the one most often skipped, because it requires holding two sections in view at once. Four checks:
- Every activity described in the method statement has a hazard in the assessment. A method that mentions cutting blockwork against an assessment with no silica hazard is a real gap, not a formatting one.
- Every hazard in the assessment has a control that appears in the method sequence. A control that exists in the table and nowhere in the described work has not been implemented, it has been listed.
- Every item of plant and every substance named in the method appears in the assessment, with its own hazards and its COSHH reference.
- The PPE in the assessment matches the PPE in the method, and both are tied to specific tasks rather than blanket-listed.
Do these as a separate pass with no judgement involved — they are comparisons, and mixing them into a careful read is how they get missed. They are also the checks an automated review runs first, which is why a document that has been through one arrives with the mismatches already listed rather than waiting to be found.
The parts that can legitimately be reused
None of this means starting from a blank page every time, and a reviewer does not expect you to. Your standing arrangements are reusable and should be: the incident reporting route, the general PPE policy, the competence framework, the legislation your trade engages, and the standard sequence for an activity you always perform the same way.
What cannot be reused is anything that describes the project — the location and everything that follows from it, the people, the interfaces with other trades, the plant and substances for this job, and the site information you were given. Keeping those two categories visibly separate in your template is the single change that most reduces the chance of a copied document going out, because an unfilled project field looks wrong where a copied paragraph looks finished.
Keeping it specific through revisions
Site-specificity decays. A scope changes and the register does not follow; a first aider leaves; the muster point moves as the site develops. Any change to the work, the conditions or the people is a trigger to re-check the five signals above, not just to increment the revision number. How often a RAMS should be reviewed sets out the full list of triggers.
One practical habit: keep the site-specific fields visibly grouped at the front of your template — names, location details, emergency arrangements, referenced surveys — so an unfilled one is obvious rather than buried on page eleven. If your document consistently reads badly because content sits in images or unusual tables, how non-standard layouts are handled is worth a look, because a human reviewer struggles with the same things a tool does.
The test
Hand the assessment to somebody who has never been to the site and ask them to describe the job. If they can tell you what is being done, where, by whom, next to what, and what happens if somebody is hurt, it is site-specific. If they can only tell you what your trade does in general, it is not — whatever the address on the front page says. How to conduct a site-specific risk assessment walks the five-step process behind it.
This article is general guidance on UK construction documentation practice. It is not legal advice, a compliance certification or a substitute for review by your own qualified health and safety personnel, who remain responsible for deciding whether any particular document is adequate. See our disclaimer.