How geofencing works
You draw a shape on a map — a radius around a shop, a district, a set of postcodes, sometimes a building or an event ground — and the platform only enters auctions for people it believes are inside it. Everyone else is filtered out before the ad is served.
The important part is how the platform decides who is inside. It combines whatever signals the device offers: GPS where an app has permission, wifi networks and mobile masts, the IP address, and the places the person searches about or has recently been. Each of those has a different accuracy, and the platform blends them into a best guess. So a boundary is a strong preference, not a fence. Someone standing just outside it may be included; someone inside on a poor connection may be missed.
Why geofencing matters
Where a business only serves people who can physically reach it, the map is the cheapest qualifier available. A boundary drawn around a delivery zone or a catchment area removes clicks that could never have converted, which stretches a modest budget further — worth more in a small market like Nepal, where the difference between a useful month and a wasted one can be a few hundred misplaced clicks.
It also allows the message to fit the place. An ad and landing page written for one neighbourhood, naming a road the reader recognises and a phone number they will actually ring, does more work than a national message shown locally.
Where geofencing goes wrong
The biggest problem is that it is sold as more precise than it is. Vendors talk about fencing a rival’s car park or a single conference hall and retargeting everyone who stood in it. In practice the signals are too coarse and too dependent on app permissions for a boundary that small to be reliable, and much of what is sold as geofencing is ordinary radius targeting with a better name. Treat any claim of building-level accuracy with suspicion.
Two settings cause the rest of the trouble. Leaving the broad presence-or-interest option on means people who merely searched about the area are included, which quietly widens the fence. And a boundary drawn tightly enough to be genuinely precise often starves the campaign of data, so automated bidding never settles.
How to act on it
Draw the boundary around a real commercial fact — the area you deliver to, the distance customers will travel, the districts your staff cover — and not around a landmark that sounded clever in a meeting. Set the targeting to people present in the area when the business is walk-in only, and check the location report to see where impressions actually landed, since that is the only honest test of whether the fence held. Add exclusions for the places you cannot serve rather than trying to shrink the shape further. If organic map visibility is the real gap, that work sits in local SEO instead, and no amount of fencing will substitute for it.
