Why two sites place your IP in different cities

Same IP, two different cities

It's a common experience: one site places you in your own city, and the next one puts you three hundred kilometers away. This isn't a one-off glitch in either service. Both are querying an IP geolocation database, and those databases don't always get the city right, even though they usually get the country right. The question almost nobody asks before picking one of these data providers is: what accuracy does it actually claim?

We reviewed the official websites of six providers in this space — MaxMind, DB-IP, IP2Location, IPinfo, Ipregistry, and ipstack — to see what each one says about its own accuracy. The result: only one of the six publishes a measured figure. The other five describe their product with adjectives, without a percentage or a geographic scope attached to it.

Country and city are not the same question

Almost any IP geolocation database gets the country right nearly all the time: mapping IP ranges to countries is well documented and changes rarely. The city is a different problem. It depends on how the internet access provider distributes those IPs among its customers, on whether corporate or mobile-carrier traffic gets rerouted, and on how often the underlying map gets updated. When a site talks about "accuracy" without saying which of the two questions it means, the good number (country) can end up standing in for the bad one (city).

The only provider that shows a number

MaxMind is the only one of the six that publishes a country-by-country breakdown of the percentage of IP addresses it places correctly within a 50 km radius at the city level. The figures it shows span a wide range: 100% accuracy in Andorra, 61% in the United States, 52% in Algeria. The provider itself adds a relevant caveat: it does not guarantee maintaining that level of accuracy over time. In other words, even the one provider that does measure and publish its accuracy doesn't promise that the next data point will behave like the historical one.

That 61% figure for the United States — the largest, best-documented market in the sector — is the number that explains why two services can disagree on the city almost four times out of ten, even in the country where the most data is available.

What the other five say

It's worth being precise about what each row actually means: a provider not publishing a percentage says nothing about whether its data is good or bad. It only says that the figure isn't available to compare from the outside.

"99.9% uptime" is not accuracy

Ipstack is the case where it's easiest to get confused. Instead of an accuracy figure, its site publishes "99.9% uptime." These are two completely different things: uptime measures whether the service answers when you query it; accuracy measures whether the answer it gives is correct. A service can be available 99.9% of the time and still get the city wrong half the time — both facts coexist without contradiction. It's worth reading carefully which percentage is being shown on any given page, because the format — a number with a lot of nines — looks the same in both cases.

Free tiers, and their fine print

All six providers offer some free tier, but with conditions worth reading before integrating anything into a website:

• MaxMind GeoLite2: free under a Creative Commons Attribution-ShareAlike 4.0 license, requires crediting MaxMind, and the provider itself warns that it "does not reliably identify any geographic level more precise than postal code."
• DB-IP Lite (country, city, and ASN): Creative Commons Attribution 4.0, requires citing DB-IP and, when used on a website, linking back to it wherever the results are shown.
• IP2Location LITE: free for personal or commercial use, with mandatory attribution; the provider itself acknowledges it as "limited accuracy." Its site names no Creative Commons license.
• IPinfo Lite: free with unlimited requests, but only at the country level plus ASN — no city, no proxy, VPN, or Tor detection. Commercial use requires visible attribution.
• Ipregistry: 20,000 free credits that don't expire, with every field included — more generous than typical, where free tiers usually strip out fields. No attribution required.
• ipstack: 100 requests per month, location module only. No timezone, currency, or VPN detection, which is only available on its most expensive plan.

The pattern that repeats is that the free tier is almost always limited to country level, and that several of these licenses — MaxMind, DB-IP, IP2Location, IPinfo — require displaying visible attribution on the very site where the data is used. That's a legal obligation of the license agreement, not a minor detail.

Other differences worth checking

Update frequency varies: IP2Location, IPinfo, and Ipregistry update daily, MaxMind every business day, and DB-IP once a month for its Lite databases. Ipstack does not publish this figure at all. All six providers cover IPv6, so that point doesn't distinguish between them.

Redistributing the data to third parties is prohibited without written permission at IPinfo and at Ipregistry; the latter also specifies that the data "is licensed, not sold" — a distinction that matters if the plan is to resell it or bundle it into another product.

Entry-level pricing, from lowest to highest: ipstack $12.99/month, Ipregistry $29/month, IPinfo $31/month, MaxMind $34.83/month, and IP2Location $49/month. DB-IP publishes its price in euros, €8.29/month, with no dollar equivalent on its own site.

What to ask before choosing one

With this laid out, the questions that have an objective answer — and that are worth asking any provider before integrating it — are:

• Is the accuracy figure it publishes about country or about city? These are different problems, and the second one is almost always left unsaid.
• Does that number come with a geographic scope, or is it an adjective with no figure attached?
• Is the "uptime" it shows service availability or data accuracy? It's worth not assuming that a high percentage answers the question you actually care about.
• If using the free tier, what license does it require, and what attribution has to appear on the site?
• How often is the database updated, and is that published, or does it have to be assumed?

None of these questions has one answer that holds for all six providers equally. Each provider's page on this site collects what it publishes, and this comparison doesn't point to a winner — it lays out the elements to decide based on what each project actually needs.

Tools to check it with your own IP

To see what location a service is assigning you right now, these tools on the site let you check directly: what IP and location is currently being shown for you, look up the location of a specific IP, and check whether an IP is flagged as a proxy or VPN. To check who owns a range of addresses, there's the autonomous system names directory, and the full list of providers in this comparison is available in the IP data services section.