Geolocation stopped being an exotic feature years ago. Today it is integrated into last-mile delivery, fleet control, local marketing, and even in the way a website decides which currency to show you. But between the theory in the manuals and the practice, there is a distance that can only be appreciated when the GPS fails in an alley, when the browser asks for permission and the user says no, or when the API returns a position with an 800-meter error.
This article does not explain what geolocation is. It gathers concrete implementation experiences, recurring mistakes, and limits that are worth knowing before promising millimeter precision to a client or a boss.
Where it is really used: four scenarios with details that matter
1. Urban delivery: the geofence that doesn’t match reality
A common case in courier companies: a service polygon is defined with coordinates and a geofence is configured to accept orders only within that area. The problem appears when the polygon is drawn on an outdated satellite map and cuts in half a street that actually belongs to the neighboring municipality. The result is rejected orders at the border and angry customers.
The practical lesson: geofences must be validated with real postal address data, not just with lines on the map. According to documentation from map providers, administrative boundaries change and base layers have update lags that can exceed a year in peri-urban areas.
2. Local marketing: the radius of influence that nobody measures
Advertising campaigns with geographic targeting allow you to choose a radius around a store. A frequent mistake is assuming that a 5 km radius is equivalent to “people who can come walking.” In a city with a highway or a river in the middle, that radius includes neighborhoods with terrible accessibility. In practice, it is advisable to combine the radius with traffic data and with the real location of active users, which advertising platforms do offer.
This does not work the same for home service businesses as for retail: in the first case, the business pays for the travel and the radius must reflect driving times, not straight-line distances.
3. Fleet control and assistance: the report that arrives late
Tracking systems send positions periodically. In areas with irregular mobile coverage, the device stores the positions and sends them in a batch when it regains signal. Whoever consumes that data must be prepared to receive a burst of points with old timestamps. If the control panel plots those points as if they were current, the fleet manager makes decisions based on outdated information.
A detail that is only learned through real use: the device battery conditions the sending frequency. Teams configured to report every 30 seconds drain the vehicle or terminal battery on long workdays. The usual approach is to alternate modes: high frequency when moving, low frequency during prolonged stops.
4. Web and apps: IP geolocation versus browser GPS
When a website requests location, there are two very different paths. The browser’s geolocation API uses GPS, wifi, and cell towers, and requires explicit user permission. IP geolocation does not ask for permission and is much less accurate: it can place the visitor in the city of the internet provider, dozens or hundreds of kilometers from their real position.
In ecommerce projects, using IP to show “your nearest store” works as an approximation, but it fails with users on corporate networks or VPNs. The usual recommendation is to offer a manual selector as a fallback and never block access to purchase due to erroneous geographic detection.
Recurring mistakes and how to avoid them
- Confusing precision with accuracy. GPS can indicate a 10-meter radius and be off by 40 due to signal bouncing between buildings. In narrow streets, the position jumps from one sidewalk to the other without the user moving.
- Not handling permission denial. Many developments assume the user will accept. If they refuse, the application must continue working with a default location or with manual input.
- Storing coordinates without a reference system. Mixing WGS84 with other coordinate systems produces deviations of hundreds of meters. You must document which system each data layer uses.
- Ignoring privacy. Storing the position history of users or employees requires a legal basis and transparency. According to the data protection regulations applicable in the European Union, continuous tracking of workers requires justification and proportionality.
- Not testing indoors. Inside industrial warehouses, shopping centers, or underground parking lots, GPS simply does not work. If the use case depends on indoor positions, you must resort to beacons or wifi, with other infrastructure and other costs.
What geolocation does not solve
It is important to be honest: geolocation does not replace a well-maintained address system. An order with an incomplete address remains a problem even if the customer has shared their exact position. The coordinate of the delivery point does not indicate the building entrance or the floor.
Nor does it solve attribution in advertising: knowing that a user was near a store does not prove that they visited it. Reliable measurement of physical visits requires other techniques, such as surveys or panels, and even then the data are estimates.
And it is not a security solution on its own. Devices can lose signal, be turned off, or be tampered with. Any location system for protection purposes must be combined with other signals and with human response protocols.
How to start without stumbling: a practical sequence
- Define the use case and its tolerance for error. A hiking app needs meter-level precision; a local marketing campaign can work with kilometers.
- Choose the appropriate position source. Device GPS for high precision, IP for frictionless approximation, mobile network data as a backup.
- Design the permission flow. Explain to the user why you need the location before launching the browser or operating system dialog.
- Test under real conditions. Walk through narrow streets, enter a shopping center, simulate low coverage. Emulators do not reproduce these failures.
- Record data quality. Store the reported accuracy alongside each position. That field allows you to filter doubtful points after the fact.
- Review privacy from day one. Define what is stored, for how long, and for what purpose.
A note on costs and provider dependency
Map and geocoding APIs have terms of use and prices that change. A project that works today with a free quota can become expensive as request volume grows. It is advisable to review the provider’s official documentation and plan for a plan B, such as caching geocodings of recurring addresses to avoid repeating identical queries.
In summary: geolocation is a powerful tool with concrete limits. The projects that work best are those that assume those limits from the design stage, offer manual alternatives, and treat position as just another piece of data, with its margin of error and its privacy policy.