Geofence alerts in TeamMap: setup and tuning

A geofence is a shape on a map that notices when someone crosses it. Set it up well and a dispatcher learns a guard left the property before the client does. Set it up badly and the alerts run all night unread.

What are geofences actually good for?

Three jobs, and they want different settings.

Watching a perimeter

• A guard leaves the area they are posted to
• Someone enters a zone they should stay out of
• Construction site boundaries after hours
• Warehouse and distribution yard perimeters

Verifying a patrol

• Confirming a guard reached every zone the route calls for
• Time spent in a specific area
• Parking structure coverage, level by level
• Building-specific assignments on a multi-building campus

Keeping a lone worker safe

• Entry into a hazardous area
• Zone monitoring for someone working alone
• Evacuation verification: who is still inside
• Site arrival and departure logging

The third one is a different product problem from the first two, and geofencing alone does not solve it. Pair it with scheduled welfare check-ins, because a zone tells you where a guard is, not whether they are all right.

How do you create one?

Open the TeamMap web dashboard, go to Geofences, and choose Add Geofence.

Drawing the boundary

Shape	Best For	Drawing Method
Circle	A single point with a radius around it	Click the center, drag the radius
Rectangle	Square or rectangular areas	Click two opposite corners
Polygon	Irregular property lines	Click each vertex

Trace the real property line with the polygon tool rather than dropping a circle over the whole block. Then leave the buffer alone until you have watched a week of live crossings, because the right buffer is a function of the site, not a number from a blog.

Choosing what the zone is for

The type field files the zone under a heading and tells everyone else what it is doing there. What actually happens on a crossing comes from the zone's entry and exit actions, not from the type.

Type	Means
Inclusion zone	Somewhere guards are meant to be
Exclusion zone	Somewhere they are meant to stay out of
Notification zone	Crossings are worth telling someone about
Alert zone	Crossings matter enough to wake someone up
Tracking zone	Kept to watch movement through it, quietly

Configuring the trigger

A trigger fires on one of four events. Enter and exit are the obvious two. Dwell fires when someone stays, which is the one you want on a loading dock where the problem is not arriving but not leaving. Cross fires on the transit itself.

Each trigger can be aimed at specific people, groups or devices, and each carries its own cooldown in minutes, so a zone that a supervisor drives through eight times a day stops reporting it eight times.

Zones can also carry a schedule: a timezone, days of the week, time ranges, date ranges and exceptions. That is how a delivery yard stops alerting between 6am and 6pm without anyone remembering to switch it off.

Why do geofences produce false alerts?

Because the phone is guessing, and it is honest about guessing badly. GPS.gov puts it plainly: smartphones are typically accurate to within a 4.9 meter radius under open sky, and accuracy worsens near buildings, bridges and trees. Guard work happens almost entirely near buildings.

So a 30 meter zone next to a parking structure is not measuring a boundary. It is measuring drift. The guard sitting still at a desk crosses in and out of it all night, and by 3am the dispatcher has stopped looking.

Three settings do the tuning, and they are team-wide rather than per zone.

Setting	Default	What it does
Default zone size	100 m	How wide a new round zone starts out. Range 25 m to 5 km.
Exit margin	20 m	How far past the edge someone drifts before they count as gone. Raise it if people on a boundary keep bouncing.
Exit margin on large zones	10%	On a big zone the margin grows to this share of its width. Whichever is larger, the meters or the share, is the one used.
Confirm arrivals after	2 position updates	How many readings in a row have to come from inside before an arrival is recorded. Two or more stops a drive-past opening a visit.

Add per-zone cooldowns on top of that, and pick the trigger event you actually care about. Alerting on both enter and exit when only one matters doubles the volume for nothing.

Who gets told, and how?

Alert	Typical use	Priority
Entry	Someone entered a restricted area	High
Exit	A guard left their patrol zone	Medium
Missed checkpoint	A checkpoint on the route was skipped	Medium

The delivery paths run in parallel. In-app notification always. Push to the person who crossed plus your admins and dispatchers. A channel message where the whole team should see it. And the activity log records every crossing whether or not anyone was notified, which is the part that matters when a client asks about a night three weeks ago.

What do you do when one fires?

A guard left their zone

1. Check whether the exit was authorized: a break, an escort, an incident call.
2. Message them in their channel, or call.
3. Write down the reason.
4. Dispatch replacement coverage if the post is now empty.
5. Note the explanation against the crossing so the next reviewer is not re-investigating it.

Someone entered a restricted area

1. Identify who triggered it.
2. Check whether that entry was authorized.
3. If it was not: dispatch a response, document it as an incident, and follow your intrusion procedure.
4. If it was authorized but unexpected: log it and fix the authorization list so it stops surprising you.

A checkpoint was missed

1. Look at the guard's last known position on the map.
2. Try to reach them by message or voice note.
3. No answer: dispatch the supervisor or the nearest guard, start a welfare check, and document both.
4. If the same checkpoint keeps getting missed, the route is wrong, not the guard.

That last line is worth taking seriously. A checkpoint that is repeatedly skipped is usually placed somewhere the patrol has no reason to go, which is a route design problem covered in routes that repeat on schedule.

How should you lay out zones across a site?

In layers, from the outside in.

• Outer boundary. The site perimeter. Exit triggers a response.
• Inner zones. Buildings and areas. These track patrol completion rather than raising alarms.
• Restricted areas. The server room, the pharmacy, the cash office. Entry triggers an alert.

Size each layer for what it is measuring, not for how it looks on the map.

Use case	Sizing approach	Notes
Presence at a specific door or asset	Do not use a geofence	Use an NFC or QR checkpoint. GPS cannot resolve a doorway.
Building coverage	Footprint plus a real buffer, starting at the 100 m default	Include the entrances people actually use
Patrol zone	The full assigned area	Leave no gaps between adjacent zones
Perimeter alert	Property line plus buffer, tuned with the exit margin	Watch a week of crossings before you tighten it

The four mistakes that cause most of the noise

• The zone is too small, so GPS drift alone crosses it.
• Arrival confirmation is left at one position update, so a drive-past opens a visit.
• No cooldown on a zone people cross all day, so the same crossing alerts over and over.
• No response procedure, so alerts arrive and nothing happens. That is the one that kills the whole system.

Rolling geofencing out to a crew that has never been tracked is its own problem, and it is mostly a trust problem rather than a technical one. Rolling out GPS without losing trust covers what to tell people and when, and geofencing without the false alarms covers the operational case in more depth.