Lone worker check-in texts and the HSE guidance

HSE's guidance on lone working is a short leaflet, INDG73, last revised in March 2020, and the part that matters most for this post is one bullet point. Among the ways to keep in touch with people who work alone, it lists "pre-agreed intervals of regular contact between the lone worker and employer, using phones, radios, email etc". A lone worker check-in text is one of the simplest ways to do that: a message at the time a visit should end, a one-word reply from the worker, and a clear rule for what happens when the reply doesn't come.

The rule for what happens next is where most of the thought has to go. A text is cheap and quick, and it is only as good as the procedure behind it.

What HSE asks of employers

INDG73 defines a lone worker as "someone who works by themselves without close or direct supervision". Its list runs from people alone in shops, petrol stations and leisure centres to cleaners and security staff working outside normal hours, care workers visiting people's homes, engineers, estate agents, delivery drivers and agricultural workers. It also covers volunteers working on their own for charities, and the employer's duty extends to "any contractors, volunteers or self-employed people" it engages.

HSE is clear that working alone is often fine: "It will often be safe to work alone. However, the law requires you to think about and deal with any health and safety risks before people are allowed to do so." There's "no legal requirement to conduct a specific, separate risk assessment for lone workers", but the risks have to be in the general risk assessment, and an employer with five or more workers must write down what it found. Among the things the leaflet says to consider are violence, the medical suitability of the person to work alone, and "having systems in place to keep in touch with them and respond to any incident".

On monitoring, INDG73 says a system "needs to be embedded into an organisation so it is well understood by workers", and that "effective means of communication are essential". Its examples include supervisors visiting, the pre-agreed contact intervals, alarm devices that can be "operated manually or automatically", and a system to make sure a lone worker has got back to base or home "once their work is completed". It also says to "regularly test technical solutions and all emergency procedures", and that emergency procedures should include guidance on "how and when lone workers can contact their employer".

Two smaller points in the leaflet bear directly on texts. Contact arrangements should bear in mind "the worker's understanding of English", so a check-in text in plain words (or in the worker's own language) is part of the design. And HSE's description of the risk is careful: lone working "does not automatically imply a higher risk of violence, but it does make workers more vulnerable", because nobody is nearby to help.

What the violence figures show

HSE publishes violence at work statistics from the Crime Survey for England and Wales. Its latest release, published on 20 November 2025, estimates 689,000 incidents of violence at work in 2024/25, made up of 370,000 assaults and 319,000 threats, against 642,000 incidents in 2023/24. About 329,000 working adults experienced at least one incident, or 1.2% of people in work. HSE says the number "fluctuates on an annual basis with no clear trend".

The risk is uneven. Over 2023/24 and 2024/25 combined, 6.8% of people in protective service occupations were victims of violence at work, which HSE notes is about six times the 1.2% average. Health and social care associate professionals came next at 2.7%, then transport and mobile machine drivers and operatives at 2.4%, sales occupations at 2.2% and other managers and proprietors at 2.0%. Those are survey estimates with wide margins (the protective services figure has a 95% confidence interval of 3.7% to 9.9%), but the order matches INDG73's list of risk factors: authority over the public, handling cash, late or early hours, alcohol and drugs.

Other details from the survey help with the design of a check-in. In 54% of incidents the offender was a stranger. Offenders were thought to be under the influence of alcohol in 40% of threats and 21% of assaults. And 38% of assaults resulted in a physical injury. A second source, the Labour Force Survey, puts the number of workers in Great Britain injured by physical violence at work at around 61,000 a year (averaged over 2022/23 to 2024/25), about 10% of all non-fatal workplace injuries. A check-in system's value is in shortening the time before anyone knows something has gone wrong.

How lone worker check-in texts work

The first part is a known finish time. The worker, or the system that books their visits, records where they're going and when they expect to be done. That record stays in the employer's system. The text never needs to carry the address.

The second is the check-in itself. At the expected finish time, a text asks the worker to confirm they're safe, and a one-word reply closes the visit:

Hartwell Care: are you finished and safe after your 14:00 visit? Reply OK. If you need more time, reply LATE.

A LATE reply can push the next check back by an agreed amount and send a short confirmation:

Hartwell Care: thanks, we'll check in again at 15:30. Reply OK when you're done.

The third part is escalation, and it's the part that has to be written down and tested. If no reply arrives within the agreed window (the employer decides how long, based on the risk assessment), a second text goes:

Hartwell Care: we haven't heard from you after your 14:00 visit. Please reply OK now. If we don't hear back in 10 minutes, your supervisor will call you.

If that one goes unanswered too, the system texts and calls the duty supervisor, who follows the procedure: call the worker, call the place they were visiting, and decide when to involve the police. The texts prompt people to act, and the decisions stay with the supervisor.

A few things make this work in practice. The workers have to know the procedure and the exact replies, which is the "embedded" and "well understood" part of INDG73. The words OK and LATE, or whatever you choose, should be short and hard to get wrong on a phone in a hurry. And the system should be tested on a quiet day, end to end, including the supervisor's phone, which is what HSE's advice on testing technical solutions and emergency procedures comes down to.

What a text can't do

A check-in text is not an alarm, and nobody should treat it as one. A worker in danger needs to call 999, and some employers give higher-risk staff a dedicated alarm device as well (HSE mentions devices that raise the alarm "manually or automatically"). Texts depend on mobile coverage and can arrive late; our post on rural not-spots covers the coverage side. That's why the escalation rule has to treat silence as a reason to act, whatever the cause.

It also matters what "delivered" means. A delivery report says the network handed the text to the handset, which is useful (a failed text to a worker's phone is itself a reason to escalate), but it doesn't mean the worker read it. Only the reply means that. Our post on what delivered means goes into the detail.

Building it on an SMS API

The scheduling and escalation logic belongs in the employer's own system: a rota or visit booking tool, or a small service written for the purpose. CaptiFi Connect does the messaging. The check-in text is a POST /messages call with a schedule_at time set to the visit's expected end, so it goes even if nobody is at a desk. Replies arrive as a message.received webhook, signed with HMAC SHA-256, which the system matches to the open visit by the worker's number. A message.failed event, meaning the text couldn't be delivered, can trigger the escalation straight away.

The texts should come from a dedicated UK mobile number so that workers can reply, and every reply also shows up in the shared two-way inbox where a supervisor can see the conversation. A keyword can send an automatic acknowledgement to an OK reply, though the matching and the escalation still need the webhook. If the employer cancels a scheduled check-in (the visit ended early and the worker checked in from the app instead), the credits come back.

One setting to watch: staff numbers and customer numbers share one opt-out list per account. A worker who replies STOP to a check-in text, by accident or not, stops all texts to their number until they text START. Keep staff alerts on their own account if you also send marketing, and make the check-in system treat a STOP like no reply. Our staff alerts page and the post on shift alerts by text cover the rest of the staff texting picture.

Common questions

Is there a legal requirement to monitor lone workers?

HSE says employers must assess and control the risks of lone working as part of their general risk assessment, and put in place ways to keep in touch and respond to incidents. INDG73 doesn't mandate one method; supervisor visits, regular contact at agreed intervals and alarm devices are all listed.

How often should a lone worker check in?

HSE leaves the interval to the employer's risk assessment, saying the higher the risk, the greater the level of supervision needed. One simple design is a check at the end of each visit or shift, with shorter intervals for higher-risk work.

What happens if a lone worker doesn't reply to a check-in text?

That should be set out in a written procedure before the system goes live. A typical one sends a second text, then alerts a supervisor to call the worker and the place they were visiting, and sets the point at which the police are called.

Can a lone worker check-in system run on text messages?

Yes, as part of the contact arrangements HSE describes, with the scheduling and escalation run by the employer's system. Texts aren't an emergency alarm, so workers still need to call 999 in danger, and some will also need a dedicated alarm device.