Places and Geofencing

A place card shows the name you gave it, the radius in metres, which children it applies to, and whether arrival alerts, departure alerts, or both are switched on. Underneath it lists the recent comings and goings, so you can see at a glance that School has been triggering at roughly the same time every morning and that nothing odd is happening at the edges.

On the map, places are drawn as soft circles rather than points. That matters, because it makes the tool honest about what it is doing. A phone is inside the circle or it is not, and the circle you can see is exactly the one the app is testing against. Nothing about geofencing is hidden behind a slider labelled “sensitivity”.

Places also feed everything else. The live map uses their names instead of street addresses, the location history timeline labels stops with them, and alerts read as sentences a human wants to receive. “Arrived at Football, 17:32” is a message you can act on. A pair of coordinates is not.

• Name, radius, and the children it applies to.
• Arrive and leave alerts, switched independently.
• Recent entries and exits, with times.
• Quiet hours, so the place stays silent overnight.

The first mechanism is Android’s own geofence service. The child app registers your circles with the operating system, which watches for crossings using the same low power location machinery the phone already runs for other apps. Because Android does this work centrally, it costs far less battery than an app waking up every minute to check for itself.

The second is our own check. Every position the phone reports is compared against your circles on arrival at the server, which catches crossings the system service missed because the phone was asleep, offline, or in a bad signal pocket. Two independent paths means fewer silent failures, and it is the reason an alert sometimes arrives a minute or two after the moment itself rather than not at all.

Then there is confirmation. A crossing is not announced the instant a single reading lands inside the circle. The app waits for the phone to still be inside a short time later, which is what stops a wobbling indoor position from firing an arrival, then a departure, then another arrival while your child has not moved at all. That short wait is the single most important piece of engineering in this feature.

Grouping does the rest. If several crossings land within a few minutes, they arrive as one notification rather than a burst, and repeated crossings at the same place inside a short window are collapsed into a single line in your alerts. The design goal is a small number of messages you will actually read.

Fifty metres is the default for a reason. It comfortably covers a house and its garden, or a school building and the yard around it, while still being specific enough that the alert means something. At that size a normal indoor position error does not push the phone in and out of the circle, so you get one clean arrival in the morning and one clean departure in the afternoon.

Draw a fifteen metre circle and you have built an alert generator. Indoors the phone’s estimate can drift by tens of metres with no movement at all, so the position keeps crossing your tiny boundary and the app dutifully reports it. You will mute the notifications within a week, which means the alert you actually wanted is the one you miss.

Going too large has a quieter cost. A three hundred metre circle around a school may include the shopping parade, the park and half a bus route, so “arrived at School” stops being true. For big sites, sprawling campuses, and blocks of flats, somewhere between a hundred and a hundred and fifty metres is usually the honest compromise. In the countryside, where positions lean on cell towers, larger is simply necessary.

Start with Home, and turn on departure alerts only. Knowing that a child left the house at 07:40 on a school morning is useful. Being told they arrived home is usually redundant, since you are often there. One switch, one message a day, no noise.

School comes next, with arrival on and departure on. That pair answers the two questions that account for most parental map checking: did they get there, and have they set off home. Set a fifty metre circle centred on the building rather than the gate, so a crowded yard still counts as inside.

The third is the one that changes the relationship, and it should be a place your child likes: a grandparent’s house, a club, a friend’s street. Turning on an arrival alert there says the tool is about knowing everyone got somewhere safely, not about catching anybody out. Add quiet hours to anything that does not need to wake you, and set alerts per child, because a fifteen year old rarely needs the same coverage as a nine year old.

It does not lock anything. A geofence is a notification, not a fence. Nothing stops a child leaving a place, no app is blocked when they do, and no door is bolted. If you want a tool that punishes movement automatically, this is not it, and we would argue that tool does not really exist in a form worth trusting.

It does not run in secret. The child app lists every place you have saved, by name, along with which alerts are switched on. Your child can see that Grandma’s has an arrival alert exactly as clearly as you can, which is the whole point.

It does not guarantee instant delivery. A phone that is off, out of coverage, or out of battery cannot cross anything, and an alert may land a few minutes late when signal is poor. It also does not detect people, only phones, so a phone left at home stays at home in the eyes of every geofencing system ever built.

Finally, it does not comment. There is no scoring of how often a place is visited, no flagging of a neighbourhood as risky, and no attempt to interpret what an unusual afternoon means. That interpretation belongs to you, ideally in a conversation rather than in a notification.