Work That Should Not Be Done Alone

Most of this collection is about arranging a response. This note is about the cases where a response arrives too late by definition, and the only control is not being alone.

When work that should not be done alone also depends on reliable work records, productivity reporting platform can support time, attendance and workload review without being treated as the emergency response itself. Compare any workflow with HSE lone-working guidance and keep alarm ownership, escalation and dispatch responsibilities explicit.

The test

Would a foreseeable incident incapacitate the person immediately?

If yes, a device summons help to somebody who is already beyond helping themselves, and the time to arrival is the whole of the outcome.

Which is not a control. It is a response, and the assessment should say which it is.

The second question: could a colleague present have prevented it, or reduced it materially?

Where both answers are yes, the task should not be done alone, and writing that down is the most valuable output an assessment produces.

The tasks where guidance is usually explicit

Confined space entry. The hazard incapacitates without warning, and rescue by an untrained person is the commonest way a single fatality becomes two. Guidance in most jurisdictions requires a trained attendant outside.

Live electrical work. Incapacity is immediate and the rescuer needs to isolate before touching.

Work at height in many configurations, particularly where a fall arrest system leaves somebody suspended — suspension trauma has a short timescale and requires rescue rather than an ambulance call.

Tree felling and chainsaw work, where industry guidance in several countries specifies a second person within sight or a defined contact interval.

Diving and work over water.

Certain patient handling, where equipment is rated for two operators.

Where guidance is explicit, the argument is over. The task is either done in pairs or it is not done.

The tasks where it is a judgement

Handling cash in quantity.

Visits to addresses with a history of violence.

Night work in premises with a break-in history.

Machinery with an entanglement hazard.

Here the assessment has to decide, and the deciding factor is the same test: how quickly would harm become irreversible, and would presence have changed it.

The conversation about cost

Two-worker operation roughly doubles the labour for that task.

Which is the real objection, and pretending otherwise weakens the case.

The honest framing: this is the cost of doing the task at all. If it cannot be afforded, the choice is to change the task or to stop doing it, not to do it alone with a device.

Where a service has priced work without the pairing requirement, the price was wrong — which is uncomfortable and is a commercial problem rather than a safety one.

Alternatives to pairing

Change the task so the hazard disappears: remote inspection, a different tool, a design change.

Move it to staffed hours, which costs scheduling flexibility rather than wages.

Bring the work to a controlled environment rather than doing it in the field.

Use a contractor equipped for it, which transfers the task rather than the duty and needs the contract to say so.

Each of these is cheaper than pairing and more effective than a device, and they are the middle of the hierarchy that gets skipped.

When the staffing genuinely does not exist

This happens, particularly in small organisations and out of hours.

The answer is not to proceed quietly.

Record the finding, record the interim controls, and record a date by which it will be resolved.

Interim controls might be: the task waits until two people are available, it is done during staffed hours, or it is not done.

"Proceeded alone with a device" is not an interim control, and an assessment recording it as one is the document that appears in an enforcement file.

The pressure that erodes it

A two-worker requirement is the first thing dropped when somebody calls in sick.

Which means the rule has to be structural rather than discretionary: the scheduling system should not permit a flagged task to be assigned to one person, and the escalation when it cannot be staffed should go to a manager rather than to the worker's judgement at the door.

Because the worker's judgement at the door is made under every pressure the organisation has created, and asking them to carry that decision alone is itself a failure of the arrangement.

Reviewing the list

Annually, and after any incident.

Because tasks change, equipment changes, and a pairing requirement based on a hoist that has since been replaced may no longer apply.

And because the list is the place where the organisation's actual position is recorded — what it will and will not ask one person to do alone.

That list, more than any device, is what an inspector will read first.

Writing the list so it survives

Name the task, not the role. "Entry to the pump chamber" rather than "maintenance work".

State the requirement precisely: two people, or one person with a trained attendant outside, or daylight only.

Say who may authorise an exception, and make that somebody senior enough that asking is a real step.

And publish it where the schedulers see it, because they are the people who will otherwise breach it without knowing.

The recurring argument

"We have always done it alone and nothing has happened."

Which is true and is not evidence, since the frequency of these events is low and the consequence is severe — the combination under which experience is least informative.

The honest answer is that the absence of an incident over ten years tells you little about the next one, and that the assessment is about consequence rather than about frequency alone.