A client emails on a Monday: “Nobody checked the loading dock on Saturday night. Prove otherwise.”
What you send back is the whole argument for understanding the difference between these three things. A GPS trail, a geofence report and a checkpoint log answer that email with completely different levels of authority — and a lot of security companies discover which is which on the Monday itself.
One sentence each
GPS is a coordinate with an error bar. The phone’s receiver computes a position from satellite signals and returns latitude, longitude and — the field almost nobody looks at — an estimated accuracy: the radius the device is probably inside.
A geofence is a rule applied to that coordinate. A circle or polygon on a map plus a condition: alert when the position enters, leaves, or when someone tries to clock in from outside. The fence measures nothing. It only judges whatever the GPS hands it.
A checkpoint is physical proof of presence at one named place. A QR code or NFC tag mounted on the loading dock. Recording it requires standing in front of it. No satellites, no polygons.
Treat them as one feature and you end up believing a geofence demonstrates a patrol. It does not, and the reason is physics rather than software.
How wrong GPS gets
Under open sky, the US government states that GPS-enabled smartphones are typically accurate to within a 4.9 m (16 ft.) radius (gps.gov). That is the best case: an open lot, clear view of the sky, nothing overhead.
The same source lists what degrades it: signal blockage from buildings, bridges and trees, and signals reflected off buildings or walls — multipath. Read that list again as a description of guard posts, because that is what it is.
- Indoors. Inside a warehouse, a mall or an office tower the direct satellite signal arrives weak or not at all. The phone falls back on cell and Wi-Fi positioning, and the error radius grows from metres to tens or hundreds of metres.
- Underground. Below a concrete slab there is usually no usable fix at all, and the last known position may be the entrance ramp — timestamped as though it were current.
- Urban canyons. The most deceptive case. Between tall buildings the direct signal is blocked and a reflected one arrives instead. The receiver produces a position that looks entirely legitimate — decimal places, timestamp, everything — and places the officer across the street or a block over. It does not fail loudly. It is confidently wrong.
That is why a geofence alone cannot prove a patrol. If the route covers seven points across a seven-acre campus, a fence drawn around the campus is satisfied just as well by an officer sitting in the guard shack. Draw the fence tight enough to be demanding and GPS error starts firing false exits: the system reports the officer left the property while the officer is standing in a basement corridor.
What each one is worth in a dispute
A GPS trail proves movement and rough coverage. It answers “was the patrol phone on the property during the shift?” and lets you reconstruct a route. It does not answer “did the officer enter the chemical store?”, because the chemical store and the aisle beside it both fit inside the error radius.
A geofence proves a rule, not a fact. Its value is timeliness: an alert the same shift, not a discovery a month later. It is an excellent supervision tool and mediocre evidence, because its quality is inherited whole from the GPS underneath. A clean geofence report shows that nothing tripped an alarm. It does not show where a person stood.
A checkpoint proves presence at a named place. It is the only one of the three that answers the client’s actual sentence — “the loading dock, Saturday night, 02:14” — and the only one a non-technical client understands without a lesson. Its limit is the mirror image: it says nothing about the minutes between points. That is where the GPS trail earns its keep.
Together they tell a story none of them tells alone. Checkpoints fix the facts, GPS connects them, the geofence raises its hand while the shift is still running.
Three things that change how the report reads
Estimated accuracy is data, not decoration. A fix with a three-metre radius and a fix with a hundred-and-twenty-metre radius are not the same evidence, even though both render as a dot on a map. Use it when you review. If a vendor shows you a map full of dots and never mentions the margin those dots were computed with, that is the question to ask in the demo.
The scan time comes from the server. A checkpoint scan does not travel with the phone’s clock; it is stamped on arrival. If the route crossed a dead zone and the data uploaded on the way out, the recorded time is the upload. Nothing is invalidated, but say it before you argue minutes with a client. A practical fix is to bracket the dead zone with a checkpoint at the entry and another at the exit.
Know what survives without a connection and what does not. The offline queue covers incidents, visitor entries and the start and end of a tour, and those keep the time the event actually happened. Attendance behaves differently: a clock-in needs a live connection at the moment it is taken, and the app does not keep it on the device for later. That is a routing decision, not a footnote, when a post sits three levels underground.
Building it so the physics stops fighting you
- Put checkpoints wherever the contract promises presence. Loading dock, substation, roof access, fire stairs. One point per specific promise — not one every fifty feet because more felt safer.
- Size the geofence to the property, not to the point. Its job is catching departures and out-of-place clock-ins, not surveying. A fence tuned to the metre generates false alerts, supervisors learn to dismiss them, and an alert nobody reads is worse than no alert at all.
- Use the GPS trail as narrative between points, knowing there will be gaps indoors and underground. A gap you explained in advance is fine. A gap the client finds first is not.
- Write the unscannable-checkpoint rule into the post orders. The officer still walks to the location, still observes, still files a written note saying what was wrong — code missing, area locked, tag gone. Without that note, an unscanned point looks identical whether the officer was thorough or asleep in the truck.
- Give the client a window. Most disputes end faster when the client can look at the same record the supervisor sees, rather than waiting for a PDF that arrives sounding defensive.
How CGuardPro puts the layers together
CGuardPro supports QR and NFC checkpoints, so the proof-of-presence layer lives on the wall rather than on the map. The QR guard tour system records each point against an identified officer; GPS guard tracking supplies the trail and the geofences; time and attendance uses the same positioning layer at shift start; and the client portal is where a client checks the record without an email thread.
The rest of the operation — patrol routes, incidents, reporting — sits in patrol software and on the features page. If the open question is which checkpoint technology to mount, that comparison is in QR vs NFC vs BLE.
One line to keep
GPS says roughly where. A geofence says whether that broke a rule. A checkpoint says someone was exactly there. A patrol contract is defended with the third, explained with the first and supervised with the second — and any vendor selling all three as a single feature is starting the relationship with a shortcut. If you want to walk through a specific site, get in touch.