Fifty RAMS a week is a real number on a large project or in a compliance team covering several sites. Individually none of them is hard. Collectively they defeat people, and they defeat them in a specific and predictable way.
What actually goes wrong at volume
Not carelessness. Three things, none of which is a failure of effort:
- Habituation. After thirty near-identical scaffold RAMS, your eye stops reading the standard sections and starts recognising their shape. The one with a missing residual column looks the same at a glance as the twenty-nine that had one.
- Depletion. Careful reading is effortful, and the effort is not evenly available across a day. Documents reviewed after about the fourth hour get systematically less attention, whoever is reading them.
- Drift. Without a fixed standard written down, your own threshold moves — stricter after an incident, looser under programme pressure. Two identical documents can get different answers in the same month.
None of these is solved by trying harder, and all of them are solved, partly, by structure.
Batch by document type, not by subcontractor
Group the week's submissions by activity: all the working-at-height documents together, all the excavation, all the hot works, all the mechanical installation. Reviewing eight scaffold documents in a block means you are holding one mental model rather than switching between eight.
Batching also makes comparison possible, and comparison is the cheapest quality check there is. Seven of the eight name an anchor point rating and one does not — a difference that is invisible when the same eight documents are spread across a fortnight.
One checklist, and it does not change mid-week
A fixed fifteen-point checklist, worked in the same order every time. Print it. The value is not that it reminds you what to look for; it is that it removes the drift, because the standard lives on the page rather than in your current mood.
Where several reviewers share the load, the checklist is also the only thing making their answers comparable. The RAMS document checklist is a usable starting point; the important property is that everyone uses the same one.
Make cross-referencing its own pass
Do not try to hold the scope in your head while reading the controls. At fifty documents a week you will not, and that is where the serious misses happen — not in the sections you read, but in the relationships between them.
- Pass one, mechanical: sections present, fields populated, residual risk complete, signatures and dates.
- Pass two, relational: scope against hazards, hazards against controls, plant and substances against the register.
- Pass three, judgement: is this method appropriate for this site, alongside this programme, with these people.
Three shallow passes beat one deep one at volume, because each pass has a single question and a single question is what survives fatigue.
Write the decision down, not just the comments
For each document record what was checked, what was found, what you decided and why. Not an essay — five lines. This is the part high-volume teams skip and the part that pays for itself twice: once when the same subcontractor submits again and you can see what you told them last time, and once when somebody asks, months later, on what basis this document was accepted.
It also exposes patterns you cannot see one document at a time. If four of this week's fifty came back for the same missing section, the problem is a template, not four authors, and it is fixable once rather than four times.
Let the mechanical passes run without you
Passes one and two are comparisons. They do not require professional judgement, they require unfailing attention to detail across thirty pages — which is exactly the resource that runs out at volume.
This is what Valify is for. Every submitted document gets the same full read: sections located, fields checked, scope cross-referenced against the hazard register, substances against COSHH references, plant against the register, control language checked for the phrases that are not controls. The report comes back grouped by severity, usually within five minutes, and it is identical in thoroughness on the fiftieth document of the week and the first.
What that changes about the reviewer's day is not the amount of reading — it is what the reading is for. Instead of hunting for the gaps, you arrive with them listed and spend your attention deciding which ones matter on this site. On a fifty-document week that is the difference between a process that holds and one that quietly degrades by Thursday.
Two things it does not do: it does not decide whether a method is appropriate, and it does not know your site. Findings get overruled, and that is the system working. Understanding your report covers reading one efficiently, and where a document comes back revised, resubmitting it shows the previous counts against the new ones so you can see what the revision actually moved.
When fifty is genuinely too many
Sometimes the honest answer is that the volume cannot be reviewed properly by the people available, and no process fixes that. It is worth being able to tell the difference, because pretending otherwise is how a compliance function becomes a rubber stamp that everybody trusts.
Three signals that you are past the line. Reviews are being signed without the cross-checks being run, because there is no time. The same subcontractor's documents are approved unread because they were fine last month. Or the queue is long enough that work is starting before approval — at which point the review is no longer a control at all, it is a record being created after the fact.
The responses available, in order of preference: reduce the number of documents by consolidating packages, so one RAMS covers a phase rather than five covering fragments of it; push quality upstream by publishing your rejection criteria so fewer documents arrive incomplete; automate the mechanical passes; and only then add reviewer capacity, which is the most expensive answer and the one that quietly recreates the consistency problem across two people instead of one.
A workable week
- Submissions close at a fixed time, so review is scheduled work rather than interruption.
- Everything gets an automated first pass on arrival; anything failing the published rejection criteria goes straight back.
- What remains is batched by activity and reviewed in blocks, hardest first, in the morning.
- Judgement work is capped — a realistic ceiling is around fifteen full reviews a day, and pretending otherwise is how the standard slips.
- Decisions are recorded as you go, and the week's returns are reviewed for patterns on Friday.
The test of a high-volume process is not throughput. It is whether document fifty gets the same standard as document one — and if the answer depends on who is reading and when, the process is the thing to fix. What happens if you submit an inadequate RAMS is a reminder of what the fifty are protecting against.
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.