How do you know your security guard is actually there?

Here’s a shift that looks perfect on paper. Clock-in at 21:58 on a 22:00 start. Clock-out at 06:02. Eight hours, on time both ends, nothing to review. You approve it with the rest of the week and move on.

Now here are three things that shift could have been hiding:

The guard clocked in, drove off to run an errand, parked across town, and came back at 05:45 to clock out.
The guard sat down at the lobby desk around 1 a.m. and woke up at 5.
The guard went to check a noise at the back of the warehouse at 2:40, fell on a loading dock, and lay there alone until the day shift arrived.

The timesheet is identical in all three. That’s the problem. A clock-in and a clock-out prove two moments. They say nothing about the eight hours in between, and the eight hours in between are what your client is paying for.

Why the clock can’t answer the question

Most small guard companies prove presence the same way: the punch, maybe a GPS stamp on the punch, and the guard’s own DAR. That answers “did someone show up?” It doesn’t answer “were they there at 3 a.m.?”

Supervisor drive-bys help. But you can’t put a supervisor at every post every hour, and a drive-by only tells you about the ten minutes it covered.

The third scenario is the one that should keep an owner up at night. A guard on a lone post has nobody next to them. If they’re hurt or sick, the first person who can raise the alarm may be the relief guard, hours later. A panic button helps, and you should have one. But a panic button only works if the guard can press it.

So “is my guard actually there?” is really three questions:

Are they on the property?
Is their phone still telling me anything?
Are they OK?

Each needs its own answer, because the signal that answers one is useless for the others.

How the problem actually gets solved

On the property: a boundary, not a pin. A GPS stamp at clock-in tells you where the guard was for one second. What you want is their position checked against the site for the whole shift. The trick is drawing the boundary generously. Include the parking lot and the sidewalk, because GPS wanders indoors and a guard checking the far corner of the lot is doing the job. And alert on a sustained stretch outside, never a single stray reading, or your supervisors will learn to ignore the alerts within a week.

The phone going dark. Someone who wants to disappear for a few hours doesn’t walk out of the boundary with the app running. They close the app or switch the phone off. So silence has to be a signal too. That means the phone has to send something even when the guard is standing still, so “still” and “switched off” stop looking the same.

Are they OK? Ask them. Location can’t answer this one. A guard standing at a post, a guard asleep at it, and a guard who has collapsed all look the same on a map. The answer is old and simple: a timed check-in. If the guard hasn’t done anything deliberate for a while, ask if they’re OK. If they don’t answer, tell someone who can call or drive over.

The details matter. Location readings shouldn’t count as “doing something,” because a phone in a sleeping guard’s pocket keeps reporting. Real activity should: a checkpoint scan, a report entry, a tap on a check-in button. And the interval has to fit the post, because a prompt every few minutes trains guards to ignore it.

Afterward: a record you can show. When a client says “your guard wasn’t even there,” your word isn’t enough. You need the shift drawn on a map.

How Watchbill does it

We built this around those three questions.

Left the site. You draw one large fence around the whole property and mark it as the site boundary. While a guard is clocked in there, Watchbill checks their position against it every few minutes. Outside longer than your grace period (10 minutes by default, adjustable from 5 to 60), and the site’s supervisors and, by default, the owners get “Left the site.” The time outside adds up as off-site minutes and shows on the shift as off-site 2h 10m, right beside late and left-early. One stray GPS fix never alerts. Readings the phone itself reports as badly inaccurate are ignored. Vehicle patrols are exempt, because driving the route is the job.

Location stopped reporting. While clocked in, the Field App sends a small “still here” signal every few minutes even when the phone hasn’t moved. If nothing arrives for longer than your threshold (20 minutes by default), you get “Location stopped reporting.” That covers the phone switched off, the app closed, or location turned off. A phone lying still on a desk doesn’t trigger it.

Welfare check-ins. You turn them on per site and set the interval: 30 minutes at the least, an hour by default. Watchbill watches for the guard’s last deliberate action: a clock event, a checkpoint scan, a report or DAR entry, a patrol stop, a checklist item, or a tap on Check in (I’m OK). Location pings don’t count. When the interval passes with nothing, the guard gets a push: Are you OK? Please check in. Five minutes without an answer and the supervisors and owners get “No response to welfare check.” The prompts keep coming every interval until the guard answers. On a post with an hourly tour, every scan already counts, so the guard rarely sees a prompt.

The trail. Every shift a guard clocked into has a Trail: a map of where they were, with start and latest pins, the site boundary drawn in, a pin for every “I’m OK” check-in, and any off-site minutes.

Under all of it sits shift status. A shift nobody clocks into is marked Missed once the grace period passes. Arrivals and departures outside your tolerance show as late 42m or left early 1h 10m, and supervisors are told. Together, that’s how you spot the timesheet that only looks perfect.

What to tell your guards

Location is recorded only between clock-in and clock-out, and the app tells the guard so. On Android there’s an ongoing notification for the whole shift. Nothing is recorded off shift.

Lead with the welfare check-in, because that’s what it’s for: someone working alone at 3 a.m. gets found if something goes wrong. It also catches the sleeping guard, but you don’t need to open with that.

Then tell your good guards the part they’ll like. The trail protects them. When a client insists nobody was on site at 2 a.m., a map of the guard periodically walking the property throughout the night settles it in their favor.

What it can’t do

Better to say it up front. None of this catches a guard who leaves their phone on site with a coworker; photo-required checkpoints are the counter to that. A guard who never allowed location is invisible to the boundary check, though the Devices page shows you whose phone is reporting. And off-site minutes are an approximation for supervision, not a number to run payroll from.

The setup and screenshots live on the presence and welfare feature page. Your guards already do the work. The point is being able to prove it, and finding out fast on the night something goes wrong.

Watchbill is workforce software for independent security companies, with presence and welfare monitoring built in. Draw one boundary, turn on welfare check-ins at your loneliest post, and watch a real shift. Start your 30-day trial, no credit card.