After hours is when alarms get real

Every workplace safety plan has an org chart behind it. A supervisor who checks in. A site manager who notices someone hasn’t come back. A colleague within earshot. Between 9 and 5, Monday to Friday, that informal safety net is real.

The data says risk doesn’t keep those hours.

What a year of alarms shows

For the FY26 State of Lone Worker Safety report, we analysed every activation logged by the Duress monitoring centre between 1 July 2025 and 30 June 2026: 20,433 activations, 815 confirmed emergencies, no sampling.

Two numbers stand out when you split them by time of day:

• 22% of activations happen after hours (6pm to 6am)
• 34% of confirmed emergencies happen after hours

The same skew shows up on weekends: 17% of activations, but 25% of confirmed emergencies.

Put simply: an alarm raised after hours is roughly twice as likely to be a confirmed emergency as one raised during the working day. And this isn’t a one-off pattern. We ran the same analysis on the previous year’s data with identical rules, and the after-hours share of confirmed emergencies held at 34% in both years.

Why the skew exists

We can’t see motives in the data, but the shape of it is consistent with what frontline teams tell us. During business hours, plenty of activations are precautionary: a worker feels uneasy on a home visit, starts a safety timer, presses early because backup is cheap. After hours, the threshold is different. The situations are lonelier, escalation paths are thinner, and by the time someone presses the button, the situation has often already turned.

There’s a second factor: after hours, the alarm is more likely to be the only signal. 17.7% of all activations in FY26 were raised automatically - fall detection, missed check-ins, and expired safety timers firing when the worker couldn’t, or didn’t, press anything. Those systems don’t clock off at 5pm.

The uncomfortable implication for check-in-based safety

Many organisations still run lone worker safety on a supervisor-based model: scheduled check-ins, a buddy system, a manager who calls if something seems off. The data points at the weakness in that model. The hours when a supervisor-based system is weakest - evenings, nights, weekends - are exactly the hours when an alarm is most likely to be real.

That’s not an argument against check-ins. It’s an argument that the response side of the system needs to be as awake at 2am as it is at 2pm. In FY26, the median time from any alarm to a monitoring operator on the case was 7 seconds, and that number doesn’t vary by hour, because the monitoring centre doesn’t have an after-hours mode. That is the entire point of it.

What to do with this

If you’re responsible for a team that works evenings, nights or weekends - community health, security, utilities, transport, retail close-up shifts - three questions worth asking of your current setup:

1. Who answers at 2am? Not “who is on call” - who actually picks up, and how fast? Ask for the number.
2. What happens if the worker can’t press anything? Fall detection and missed check-in escalation exist precisely for the scenarios where the worker is not able to raise the alarm themselves.
3. Does your risk assessment weight after-hours work properly? If after-hours incidents are twice as likely to be real, the controls for those shifts deserve more than a pro-rata share of attention. Our lone worker risk assessment template has a section for exactly this.

The full dataset - response-time distribution, trigger types, year-on-year comparisons, and the methodology - is published in the State of Lone Worker Safety FY26 report. Free to cite with attribution.