GPS accuracy is how close a reported position is to the true location, usually expressed as a radius within which the real position probably falls.
When a device reports a position, it usually reports an accuracy figure with it — something like 12 metres. That number is not a guarantee. It means roughly: the true position is probably within 12 metres of this point, typically at about 68% confidence. One report in three falls outside it.
Understanding that figure is the difference between a geofence that works and one that cries wolf.
Marketing material quotes best-case numbers measured under ideal conditions. What you actually get:
| Environment | Typical accuracy |
|---|---|
| Open field, clear sky | 3–5 m |
| Suburban street, some trees | 5–15 m |
| City street with tall buildings | 20–50 m |
| Dense urban canyon | 50–200 m |
| Indoors near a window | 100 m+, often unavailable |
| Underground, deep indoors | No fix |
Modern phones sit at the better end because they combine GNSS with WiFi positioning, cell towers, and inertial sensors. A cheap 2G tracker with a small antenna sits at the worse end, and falls back to cell tower positioning when it loses satellite lock — accurate to hundreds of metres at best.
Six decimal places of longitude implies 11-centimetre resolution. That is precision, and it is free — the number of digits a device prints says nothing about whether they are correct.
A receiver reporting 41.402771, 2.174379 with an accuracy of 30 metres is
being precise and inaccurate simultaneously. The last three decimals are noise.
Five decimal places is the practical ceiling for consumer GPS; beyond that you
are recording receiver jitter. The
coordinate converter has a table of what each
decimal place is worth on the ground.
This is the practical payoff, and the rule is short: never draw a zone smaller than your accuracy figure.
A 50-metre radius with 50-metre accuracy means a stationary device produces positions on both sides of the boundary continuously. You have not built a geofence, you have built a random alert generator.
| Radius | Sensible with |
|---|---|
| 50 m | Nothing — below the consumer noise floor |
| 150 m | Good open-sky GPS, a building plus its car park |
| 500 m | Most phones and trackers, a campus or village centre |
| 1 km+ | Anything, including cell-tower positioning |
Add margin beyond the property line, and include wherever a device might sit while it counts as “arrived” — the car park, the driveway. The radius map shows what a given radius actually covers, and the geofence builder reports the real dimensions of a traced shape.
Even a perfectly sized zone will see occasional boundary crossings from a stationary device, because accuracy varies minute to minute as satellites move and conditions change. Sizing reduces the frequency; it does not reach zero.
That residual is what GPS drift describes, and why alerting has to be built on state rather than raw crossings. Geometry and software each handle part of the problem, and you need both.
Last reviewed 7 August 2026.
GPS drift is the apparent movement of a stationary device caused by errors in its reported position rather than by any real change in location.
A-GPS (Assisted GPS) speeds up position fixes by downloading satellite orbit data over the mobile network instead of waiting to receive it from the satellites themselves.
LBS positioning estimates a device's location from nearby mobile network towers rather than satellites — far less accurate than GPS, but it works indoors and needs no fix.
Geofence hysteresis is the use of different thresholds for entering and leaving a zone, so a device hovering at the boundary cannot rapidly flip between states.
Draw a zone, assign a phone or GPS tracker, and get a blip the moment it crosses the line. Free for 7 days.