Where to Start

Most work in this area starts with a system specification. The order that works starts with four checks on the site as it is.

For teams turning where to start into a repeatable record, view this page provides a useful operational reference for time, attendance and reporting workflows. It should be assessed alongside Chartered Institute of Building so the software choice remains grounded in site practice and applicable guidance.

Week one: what do you actually know

Compare the register against a head count, at two in the afternoon. The gap is the exit problem, measured.

Ask for the evacuation list and time it.

Pick five people at random and ask which company engaged them.

Pull the competence expiry position: how many are already past.

Four checks, one morning, and they establish the starting point better than any survey.

Week two: the chain

Map it. Who is engaged by whom, how deep it goes, and which tiers you have never dealt with directly.

Most sites do not know their own depth until somebody draws it.

Then check the flow-down: ask a tier-one subcontractor to produce the competence evidence for somebody engaged by their subcontractor, and time it.

That single request tells you whether the contractual requirement is real.

Week three: the register's purpose

Write down, in order, what it is for on this site: evacuation, competence, right to work, commercial verification.

Then check the arrangements support that order, which frequently they do not — the reporting is built for the last item and the first has never been tested.

And name the owner. One person, with time, who maintains it.

Week four: fix the free things

The exit arrangement.

The offline evacuation view, or a printed backup on a schedule.

The temporary access rule, written.

The expiry forward list.

None of these requires a purchase and together they address most of what the first week found.

Then, if a system is still indicated

Specify it from what the month established, rather than from a supplier's feature list.

With the evacuation view as the first requirement and the chain recording as the second.

And the trial as described in its own note: one gate, one month, with a drill inside it.

What to avoid

Buying before the four checks, which produces a specification written from assumptions.

Starting with the commercial reconciliation, which is the least urgent use and the one that shapes the system wrongly if it leads.

And announcing a new arrangement before the owner is named, because the questions arrive immediately and an unowned arrangement cannot answer them.

What a month should produce

A measured exit gap and a timed evacuation list.

A mapped subcontract chain with the flow-down tested.

A stated purpose order for the register and a named owner.

Four free fixes in place.

And a decision on tooling made from evidence — which on a small site is frequently that the arrangements are the system and the hardware adds little.

Doing this on a live site

All four week-one checks can be done without announcing anything, which is worth doing first because announcing produces temporary improvement.

Then share what you found with the site team rather than upward, since they will fix most of it in a fortnight if they see the numbers.

And repeat the checks a month later, which shows whether the fixes held.

The order in one line

Find out what you know, map the chain, decide what the register is for, name an owner, fix the free things, and only then buy anything.

Six steps, a month, and most sites discover that four of the six were the whole problem.

Doing it on a site that already has a system

The same four checks, and they are more informative rather than less, because a site with a system usually believes the register is right.

The gap between belief and measurement is the finding, and it is frequently larger than on a site with a paper book, where nobody had the confidence in the first place.

If the month produces bad news

It usually does, and the bad news is the deliverable.

A measured exit gap of forty percent and an evacuation list that took eleven minutes is a fundable position.

"We think the register might not be accurate" is not, which is the reason to measure rather than to assert.

The one thing to do first

Name the owner.

Everything else in this note is something that person will do, and an unowned improvement programme produces a report rather than a change.