What Not to Do

Each of these is common, each is understandable, and each produces a system that is worse than the absence of one.

When what not to do also depends on reliable work records, the workflow guide 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.

Buying devices before assessing the risk

Why not: the assessment determines whether devices are the right control, and for some tasks the answer is that the work should not be done alone at all.

Instead: assess, then decide, then equip.

Treating the device as the control

Why not: it summons help after the harm, which for a fall from height or a confined space is not a control.

Instead: eliminate the lone working for those tasks, and use devices where the risk is one help can still address.

Continuous tracking because the product does it

Why not: it collects a complete record of the worker's day for a purpose a schedule plus an alarm position would serve, and it is the thing that makes people leave the device in the van.

Instead: position on alarm, schedule otherwise.

An escalation tree nobody has tested

Why not: most contain at least one wrong number, and the failure only appears when it matters.

Instead: quarterly unannounced tests, at the hours the work happens.

Penalising withdrawals

Why not: it converts the most effective control available into one nobody uses.

Instead: record withdrawals as system successes and make the first one visibly fine.

Tolerating false alarms

Why not: the responder's assumption becomes "another false one", and the response time lengthens without anyone deciding.

Instead: tune the sensitivity, fix the wearing position, and measure the rate.

Office-hours monitoring for out-of-hours work

Why not: it is an arrangement that covers the hours when nothing happens.

Instead: plot when lone working actually occurs, then choose monitoring to match.

A policy attached to job titles

Why not: it misses everyone who is alone intermittently, which is the larger group and the one with no arrangement at all.

Instead: assess by situation and shift.

The common thread

Most of these substitute equipment for a decision — about staffing, scheduling, or whether the task should be done alone.

The equipment is visible, purchasable and quick. The decision is none of those, which is why the substitution keeps happening.

Two more that matter

Using the system to check up on people. The fastest way to end the workforce's cooperation, and it only has to happen once. The device goes in the glovebox and the organisation does not find out until an incident.

Treating a device as evidence that the risk was managed. In enforcement terms it is the opposite: it demonstrates the risk was identified, which makes the absence of the other controls harder to explain rather than easier.

The pattern behind the pattern

Each of these substitutes something visible for something structural.

Devices for staffing. A policy document for a tested chain. A device count for a response time.

The visible thing can be shown to a board, an auditor or a client, which is precisely why it gets chosen.

The structural thing is what protects anybody, and it produces no artefact until somebody measures it.

How to tell which you have

Three questions, answerable in an afternoon.

When was the escalation tree last tested, and what was the time?

How many withdrawals were recorded last year?

What proportion of alarms are false, and does the responder still take each one seriously?

An organisation that can answer all three has a system.

One that can answer none has an estate of devices, and the difference is invisible from the outside until the day it is not.

The organisational tells

A programme owned by procurement rather than by safety.

A business case built entirely on device count.

A policy that names a product.

A risk assessment written after the purchase.

Any of these suggests the decision was made before the analysis, which is the condition under which every failure in this note becomes likely.

Recovering from it

An organisation that already has devices and no chain is not in a bad position.

Test the tree this quarter, write the assessment, state the withdrawal policy, and start measuring response times.

None of that requires new equipment, and it converts an estate of devices into a system within a quarter.

The one that is hardest to resist

Buying because something happened somewhere else.

An incident in the sector, a story in the trade press, a coroner's finding — and the reasonable impulse is to act quickly.

Acting quickly means buying, because that is the action available in a week.

The assessment takes longer and answers whether the thing that happened elsewhere could happen here, which is the question actually being asked.

Do both, in that order, and the purchase will be better specified for the delay.