False Alarms

A lone worker system with a high false alarm rate is worse than none, because it produces confidence that nobody is actually watching.

When false alarms also depends on reliable work records, see how it works 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.

Where they come from

Fall detection triggered by putting the device down, sitting heavily, or a rough road.

Panic buttons pressed in a pocket or a bag.

Check-in timers that expire because a worker was in a meeting or a lift.

No-movement alarms from someone eating lunch.

And devices knocked against a door frame.

None of these is anyone behaving badly, which is the point: the rate is a design property.

What a high rate does

The responder's first assumption becomes "it's another false one", which is rational and is exactly the failure.

Response time lengthens invisibly: nobody decides to be slower, they just check something else first.

Workers stop reporting genuine near misses, because they are tired of being called about nothing.

And the organisation believes it has a working system, which is the most dangerous state of the three.

Measuring it honestly

Alerts raised, classified: genuine, precautionary, false.

Per worker and per device type.

A single worker generating most of them is a device fit or a training issue, not a person issue.

And a rate above a few per worker per month needs addressing before anything else in the programme.

Reducing it

Tune the sensitivity to the actual work. A care worker bending over a bed is not a fall.

Fix the wearing position, which resolves a large share of pocket activations.

Lengthen check-in intervals where the hazard allows, since a short interval buys accuracy at the cost of expirations.

Train the cancel procedure, so a worker who triggers accidentally can stand it down in seconds without a phone call.

And replace devices that misfire persistently, which is a warranty conversation rather than a tolerance.

What not to do

Do not respond by ignoring categories of alert, which is the informal version of turning the system off.

Do not penalise workers for false activations, which teaches them not to press it at all.

And do not lengthen the response time to compensate, which is the same failure formalised.

The measure that matters

Time from alert to first contact attempt, trended.

If it is rising, the false alarm rate is doing its damage regardless of what the alert log says.

And that number, not the device specification, is what the system is.

Classifying properly

Genuine: the worker needed help.

Precautionary: the worker chose to raise it and did not ultimately need help. This is a success, not a false alarm, and classifying it as one teaches people not to use the system.

Accidental: pressed in a pocket or triggered by movement.

Technical: the device misfired.

Four categories, and only the last two are the problem. Most organisations collapse the middle two, which produces a number that both overstates the problem and discourages the behaviour they want.

Where the rate should sit

There is no universal figure, and any supplier quoting one is quoting a different operation.

The practical test is whether the responder still treats each alert as real.

Ask them. They will know, and they will tell you if the question is asked without implication.

A responder who says "most of them are nothing" has told you the rate is too high, whatever the log says.

The cancel path in detail

Five seconds, on the device, with no phone call.

A confirmation to the worker that it was received.

And a record that it was cancelled and by whom, so the log distinguishes a cancelled alert from one nobody answered.

Without a cancel path, every accidental activation becomes a phone call the worker must explain, which is precisely the friction that makes people stop carrying the device.

Device fit and activation position

A large share of accidental activations come from where the device sits.

A pendant under a seatbelt, a fob in a trouser pocket, a badge holder against a car door.

Move it and the rate falls, which is a five-minute conversation rather than a technical fix.

And a worker generating many accidental alerts is telling you the device does not fit their work, which is a fitting problem and not a behaviour one.

Reviewing the rate

Monthly, by category, by worker and by device model.

A model with a persistently higher technical rate is a warranty conversation.

A worker with a persistently higher accidental rate is a fitting conversation.

And a rise across everything after a firmware update is the supplier's problem, which you will only spot if the baseline was recorded.

Telling workers the rate

Publishing it removes the embarrassment.

A worker who knows that accidental activations are common and expected will report theirs, which keeps the responder's picture accurate.

One who believes theirs was unusual will quietly not mention it, and the log becomes wrong in the direction that matters.

Say the number out loud once a quarter. It costs nothing and changes the reporting behaviour immediately.