GPS tracking for guards: policy, law and rollout

Any guard app will put a dot on a map. Whether GPS tracking makes your operation safer or makes your best officers start job hunting comes down to four decisions you make before the first phone is enrolled.

Why do GPS rollouts fail?

Almost never for technical reasons. The failure pattern is consistent, and it starts with a supervisor.

Someone discovers they can see the map. They notice a guard stopped for eleven minutes at a gas station. They mention it. Within a week every officer on the account knows the map is a scorecard, and the value of the system, which was never about gas stations, is gone. The other failure is quieter: a system installed without announcement, found by an officer reading their app permissions. Then you spend six months explaining that you were not hiding anything.

So before you enroll anything, answer four questions in writing and be ready to read the answers aloud at a shift briefing.

1. Why are you tracking? If the honest answer is "so I can see if they are working," stop. That is a supervision problem and location data will not fix it.
2. What will you do with the data? Name the uses. All of them.
3. Who can see it? A named role, not "the office."
4. What happens when a location does not match expectations? Decide the response now, while nobody is in trouble.

What does the law actually require?

More than most operators assume, and the requirements are procedural rather than substantive. In general the statutes do not stop you tracking. They stop you tracking quietly.

Connecticut is the clearest model. Under General Statutes section 31-48d, an employer engaged in any type of electronic monitoring must give prior written notice to every employee who may be affected, telling them what types of monitoring may occur, and must post that notice conspicuously where employees can read it.

The statute covers collecting information about employees' activities or communications by any means other than direct observation, with an exception where the employer has reasonable grounds to suspect illegal conduct. The Labor Commissioner levies $500, $1,000 and $3,000 for first, second and subsequent offenses.

New York takes the same notice-first approach. Civil Rights Law section 52-c requires private employers who monitor telephone, email or internet usage to give written notice upon hiring, obtain a written or electronic acknowledgment, and post the notice conspicuously. It addresses communications rather than location, but it tells you what the state considers a reasonable baseline.

California is the one to read carefully if you run vehicles. Penal Code section 637.7 bars using an electronic tracking device to determine the location or movement of a person, and makes a violation a misdemeanor. The exception is narrow: it applies where the registered owner, lessor or lessee of a vehicle has consented to the device on that vehicle, which covers a tracker in a company patrol car. For a licensed business, a violation is also grounds for revoking the license.

What should the written policy say?

Six things, in plain language, on one page. Vagueness here is what creates the suspicion you are trying to avoid.

Put it in onboarding and get the acknowledgment. Then answer the uncomfortable question honestly when a guard asks whether they can be disciplined off the map. If the answer is "not on the map alone," say that and mean it. Off-shift collection is the single largest trust variable: if the app collects while a guard is clocked out, no policy paragraph will fix how that feels.

Which GPS settings actually matter?

Three, and only three. The rest is default.

Update frequency. Battery against resolution. A thirty-second interval covers dispatch, route verification and welfare checks with room to spare. A five-second interval costs battery for detail nobody uses. Start slow, measure the drain on a real shift with your own handsets, and tune from what your dispatchers actually need.

Geofence radius. Too tight and every officer walking to their vehicle triggers an alert, which trains dispatch to dismiss alerts. Too loose and the fence means nothing. Tune it against a week of real movement rather than a floor plan. Geofencing without the false alarms covers that process in detail.

Offline behavior. This is the one that breaks in production. Positions must queue on the device when signal drops and sync when it returns, and you have to test it where guards actually are: the third basement of a parking structure, the freight core, the loading dock. An office test proves nothing.

Indoor accuracy is the honest limitation. GPS degrades badly inside steel and concrete, and a fix that reads fifty meters off is worse than no fix if somebody treats it as evidence. Where you need proof of presence rather than approximate position, use NFC checkpoints alongside location, not instead of it.

What is the data legitimately for?

Four uses. If a use is not on this list, it needs to be added to the policy before it happens, not justified afterward.

• Safety. An officer misses a welfare check and you need to know where to send someone. This is the use that justifies the whole system, and it is the one to lead with when you announce it. See welfare check-ins for lone workers.
• Dispatch. Sending the closest unit instead of the one you remembered.
• Proof of service. Answering the client's question about whether the 0300 round happened, with something better than a signature. That feeds directly into client reporting.
• Coverage analysis. Finding the corner of the property nobody has walked in three weeks, so you can fix the route rather than the officer.

What must it never be used for?

Three things, and each of them ends the program.

• Break policing. Catching a fourteen-minute break at a twelve-minute site. You will win that one and lose the account's whole crew.
• Route second-guessing. Questioning a deviation with no incident attached to it. Officers deviate for good reasons you cannot see on a map.
• Building a file. Using location history to assemble a termination case that has no other evidence in it. If the performance problem is real, it will be visible somewhere that is not a map.

How do you roll it out?

Slowly, and in this order.

1. Announce it before anything is installed. Weeks, not days, and take questions in a room rather than by email.
2. Train supervisors first, on the policy as much as the tool. The first misuse sets the norm, and it comes from a supervisor who did not know the rule.
3. Pilot with volunteers. Let the officers who like the tech find the battery problem and the fence that fires every night at 0200.
4. Enroll by account, not all at once. Each site teaches you something about radius and interval.
5. Ask at thirty days, then change something visible, so people learn that asking works.

The measure of a good rollout is not adoption. It is that six months later, an officer who does not answer a welfare check gets found quickly, and nobody on the crew resents the thing that found them.