All articles

How Accurate Is IP Geolocation? Country vs. City

Vendors advertise "99% accuracy", lookup sites show a confident dot on a map — and both can be true and misleading at once. Here are the real accuracy numbers at each level, what makes them better or worse, and how to test your own IP.

How Accurate Is IP Geolocation? Country vs. City

Accuracy is a ladder, not a number

How accurate IP geolocation is depends entirely on how closely you zoom. MaxMind — whose free database powers a large share of lookup sites — documents it roughly as: country above 99%, region or state 55–80%, city 20–75%. Street level is simply not possible from an IP.

So “is it accurate” has three honest answers at once: extremely (country), usually (region), and often not (city).

What the “99% accurate” claims measure

When a vendor advertises 99%, that figure almost always describes country-level resolution on a curated test set. It says nothing about city precision for your region, your ISP or your connection type.

And when a vendor has no confident city for a block, many products fall back to a region or country centroid — which inflates apparent coverage while telling you nothing about a real place.

What makes your own result better or worse

Accuracy isn’t random — it follows the type of connection:

  • Fixed broadband in a large city — the best case; the city is often right.
  • Suburban and rural connections — usually resolve to the ISP’s nearest aggregation hub.
  • Mobile networks — traffic exits at carrier gateways; expect region-level accuracy at best.
  • CGNAT — one shared IP for thousands of users pins everyone to the pool’s location.
  • VPNs and proxies — the result describes the exit server precisely, and you not at all.
  • Free vs commercial databases — free datasets like GeoLite2 lag commercial ones noticeably at city level.

Coordinates deserve special skepticism

The latitude and longitude in a lookup are a representative point for an address block — typically a city or region centroid with an implied radius of tens of kilometers. That’s fine for analytics and wrong as a position.

Rendering that point as a precise map pin is exactly how a database estimate turns into a false conclusion. Read it as “this area”, never “this spot”.

How to test your own IP

Run a lookup and compare three things against reality: the country (should match), the city (may be your ISP’s hub — check whether the named city is where your provider concentrates its infrastructure), and the ISP/ASN field (it tells you whose network the estimate actually describes).

Then check the thing that matters more than the city name: whether your IP’s region agrees with your browser’s timezone and language. Websites weigh that consistency far more than street-level precision.

Test your IP's geolocation now

See your IP's estimated location, ISP and timezone side by side — and whether they agree with your browser.

Look up my IP
Keep reading

Related articles