Lone working: what a risk assessment has to cover

Start by mapping honestly, because the obvious list is always too short. Lone working is not just the night security guard. It is anyone who cannot be seen or heard by a colleague while they work — and that includes people inside a busy site: a picker at the far end of a warehouse aisle, a cleaner on an empty floor of an occupied building.

Then map when, because alone-ness follows the clock more than the job description. The same role can be surrounded at noon and isolated at ten at night, and the isolated hours are the ones the plan is for. The stretches assessments most often miss:

Resist the urge to invent exotic hazards. For most lone workers the realistic list is short and ordinary: a medical event, an accident, aggression from another person. The same three things that can happen to anybody at work — your existing risk register almost certainly covers them already.

What changes is not the probability but the consequence. A fall from a stepladder in a staffed workshop is a first-aid entry; the same fall on a Sunday call-out, out of phone reach, is a person on the concrete until Monday. When nobody sees it happen, severity becomes a function of time — so assess the hazards you already assess, and re-score their severity for the lone case. Lone working is rarely a new hazard; it is a multiplier on the register you already have.

This is the question most assessments dodge, and the one that matters most. Every plan that begins “the worker calls…” assumes a conscious worker with a charged, unlocked, reachable phone. A collapse, a blow to the head, a seizure — the worst outcomes are precisely the cases where nobody presses anything.

A defensible section documents at least one path that does not depend on the worker acting. In practice that means an automatic or scheduled trigger: a device with automatic fall detection that raises the alarm itself, or a timed check-in the worker must positively cancel, so that silence becomes the signal.

This is what dedicated hardware is for. ProTechMe’s standalone GPS button is a business device with its own mobile connection that raises an alarm when the SOS is held for five seconds — or automatically, when it detects a fall — and carries two-way voice through the device itself. Whatever you deploy, write down which trigger covers the cannot-press case.

An alarm that reaches “the system” has not reached anyone. Document the chain by role: who is alerted first, who is next if nobody acknowledges, and what each responder does — go to the worker, phone them, call the emergency services themselves. Responders who improvise on the night are not a plan.

Then answer the question every responder asks first: where? An outdoor GPS position is useful on a road and nearly useless inside — a dot accurate to a few metres says little when the person is inside one of four adjacent buildings. If your lone workers are ever indoors, say how a responder narrows it down.

The Danish municipality Rudersdal Kommune shows this done properly: staff carry the GPS button for their own safety, and indoor positioning built for the deployment means other units announce which building an alarm came from. Everyone who hears it knows where to run. Alarm, plus building — that is the standard to design toward.

The wording differs by country; the expectation largely does not. Denmark’s working-environment authority expects a risk assessment to cover the work actually performed — including work performed alone. Singapore’s Workplace Safety and Health framework carries the same expectation: the assessment follows the activity, not just the premises.

The practical reading: if people in your operation are ever alone — out of sight, out of earshot — an assessment that never mentions it is incomplete. Check the specific rules where you operate, but do not expect a regulator to accept “we didn’t think of the weekend call-outs”.

A plan that has only ever been read is a document, not a plan. Test it on an actual shift, under actual conditions: the night the building is empty, the gloves the worker really wears, the basement where the signal drops, the phone on silent because it always is.

The test itself is simple. Trigger a drill alarm from the loneliest point of the shift and watch: did it reach a person, did that person know what to do, could they say where the worker was? Then write down what failed, fix it, and test again.

And review on a trigger, not a calendar. After every incident and near miss, after a change in staffing or opening hours, after a move to new premises. A lone-working section dated three years ago describes a workplace that no longer exists.

Pulled together, a defensible lone-working section is not long — most fit on a couple of pages. What makes it defensible is that it answers each of the questions above in plain sentences a new starter could follow without phoning anyone to ask:

Only one of those five lines is about equipment. Devices matter — the cannot-press case is nearly impossible to cover without one — but a button bought before the assessment is written answers a question nobody asked. Do the mapping first. Then choose hardware to close the gaps the paper shows you, and hold any vendor, ours included, to these questions.