How accurate is IP geolocation?

Why country and city accuracy differ

Country assignment mostly follows how ISPs are allocated address ranges, which changes rarely. City assignment depends on how a specific range is registered and where traffic is routed, which is far less stable — a business ISP, mobile carrier, or VPN provider can route traffic through an address registered to a different city than the connection's actual user.

Cases that commonly disagree with a device's real position

Mobile carrier networks often resolve to a regional gateway rather than the handset's actual city. VPNs and proxies intentionally place the connection at the exit node's location. Corporate networks and CDNs can anycast a range across multiple physical locations. In all of these cases, the response is describing the connection, not the person using it.

Designing your own evaluation

A meaningful accuracy check needs a stated ground truth for each test address (with consent), a sample stratified by country, network type, and IP version, a defined error metric — distance in kilometers at city level, correct/incorrect at country level — and a stated test date, since address ranges get reassigned over time. Publish the sample size and methodology alongside any percentage, or the number isn't evaluable by a reader.

What this article does not claim

This page does not publish a Watch-IP accuracy percentage, because no representative, dated evaluation has been run and published yet. Treat any accuracy claim you see elsewhere, from any vendor, the same way: ask for the sample, the ground truth, and the date.

Sources

• Watch-IP's data sources — full provider and attribution details