What an address lookup can actually answer
An IP address is not a name, a location or a person. It is a routing label: a number that tells every router between here and there which direction to send the next packet. Everything a lookup tells you is derived from that one fact — from the paperwork of who was allocated the number, and from what the network does with it in practice.
That sounds like a limitation, and it is, but the useful part is larger than people expect. Five pieces of information come out of an address, and they come from four completely different places:
- Country
- From the registry allocation and from measurement. Reliable at the level of "which country", and no further — see below for why we deliberately stop there.
- Autonomous system (AS)
- The network that announces this address to the rest of the internet. This is the single most trustworthy field in the whole result, because it is not a database claim — it is what the global routing table is doing right now.
- Organisation
- Who holds the AS: an internet provider, a mobile operator, a hosting company, a university, a government network.
- Reverse DNS (PTR)
- The name the address holder chose to publish for this address. Often the most informative line on the page, and often empty.
- Network type
- Whether this looks like a home connection, a mobile network, or a machine in a data centre. An inference, not a fact — the reasoning is spelled out below.
Where the country actually comes from, and why we do not show a city
There is no field in the internet that stores "this address is in Warsaw". Geolocation is assembled from several sources of varying quality: the country a block is registered under at its regional registry, the routing announcements that show where it is connected, latency measurements between known points, hostname conventions in reverse DNS, and data that operators publish about their own ranges.
Assembled that way, country-level answers are right the overwhelming majority of the time. City-level answers are not, and the gap between the two is far wider than the industry likes to advertise. Three ordinary situations break city geolocation completely:
- Mobile networks. A phone's traffic leaves the operator's network through a gateway that may be hundreds of kilometres from the phone. An entire country's mobile subscribers can appear to be in one district of the capital, because in routing terms they are.
- Provider backhaul. A regional ISP commonly hands out addresses from one pool across many towns. The database knows the pool, not the town.
- Stale data. Address blocks are traded and reassigned constantly. When a block moves between operators — or between countries — the databases catch up over weeks, not minutes.
So we publish the country and stop. A page that prints a street-level map next to an address is showing you the centre of a region with a confidence it does not have, and the well-known consequence is real people receiving visits at a farm in Kansas because an unlocatable address defaulted to the geographic centre of the United States. The honest answer is the coarse one.
The autonomous system is the field to trust
An autonomous system is a network with its own routing policy — a provider, a large hosting company, a bank, a content network. Each has a number, and that number is what other networks use to decide how to reach it. Numbers were originally 16-bit; that space ran out, and 32-bit AS numbers were standardised in 2012, which is why you now see values far above 65535.
What makes the AS worth more than the geolocation is its source. Country data is a claim in a database. The AS is observable: the address is inside a prefix that some network announces to its neighbours, right now, and if it stopped announcing it the address would become unreachable. There is nothing to be out of date about.
Practically, the AS answers questions the country never will. Traffic from a residential provider's AS at three in the morning behaves differently from traffic out of a hosting company's AS, and "this visitor came from a cloud provider" is a far more actionable statement than "this visitor is in Germany". If you want the registry paperwork behind the AS — who holds it, which ranges it covers, where to send a complaint — that is what the whois and RDAP record is for.
Reverse DNS: the line people skip
Reverse DNS maps an address back to a name, using the special in-addr.arpa zone
for IPv4 and ip6.arpa for IPv6. The crucial detail is who controls it: the PTR
record is published by whoever holds the address block, not by whoever owns the website running
on it. That is why the reverse name of a shared hosting server is the hosting company's, not
the site's, and why anybody can publish a forward record claiming to be
mail.big-bank.example while the reverse name gives away the truth.
Operator naming conventions make PTR unusually informative. Names in the shape of
host-89-64-12-7.dynamic.example-isp.net or
93-115-3-2.static.customer.example.net quietly tell you whether the address is
handed out dynamically or assigned to one customer, and often which city or exchange it hangs
off. An empty PTR is a signal too: on a mail server it is close to fatal, because a large share
of receiving systems refuse or heavily penalise mail from an address with no reverse name.
The verification step worth knowing is forward-confirmed reverse DNS: take the PTR name, resolve it forward, and check that it comes back to the address you started with. Anyone can claim a name in reverse; only the name's real owner can make the forward record agree. This is exactly how search-engine crawlers should be verified, and it is a check you can run yourself with a DNS lookup on the name the PTR gave you.
Network type, and the honest limits of it
We label an address as residential, mobile, hosting or unknown by combining the kind of organisation that holds the range, the reverse-DNS naming, and how the block is used. This is an inference. It is usually right, and it is wrong often enough that it should never be the sole basis for blocking anybody.
The specific traps: a company's office traffic may leave through a data-centre link and look
like hosting; a VPN or a corporate proxy makes a home user appear to be a machine in a rack; and
carrier-grade NAT puts thousands of unrelated households behind a single public address. That
last one matters more every year. Because IPv4 space ran out — the central pool was exhausted in
2011, and the European registry stopped issuing new blocks in 2019 — providers now place many
customers behind one shared address, with 100.64.0.0/10 reserved for the private
side of that arrangement. If you block a single address on a mobile network, you may be blocking
a town.
Using this in practice
- An unfamiliar address in your server log
- Look at the AS first. A scan from a hosting provider's range needs a different response from a login attempt out of a residential network in the country where your staff live.
- Checking whether a service is where it claims to be
- Resolve the domain, look up the resulting address, and compare the organisation with the one the site claims. A "local" service whose servers sit in a single low-cost hosting AS is not a scandal, but it is worth knowing.
- Mail that will not deliver
- Check the reverse name exists and matches, then check the address against the blocklists that receiving systems consult. These two account for most rejected mail from a correctly configured server.
- Diagnosing "the site is slow for me"
- Knowing the network is only half of it; measure the path with real ICMP packets before concluding anything about distance.
What this page will never tell you
It will not tell you a name, a household, a street or a device. It will not prove that an address belongs to a VPN, only that its network is the kind of network VPNs run on. And it cannot tell you what an address was doing last week: allocations change hands, dynamic addresses are reassigned nightly, and today's answer is a statement about today.
Treat the result as the beginning of an investigation rather than a verdict. The registry record explains who is accountable, the DNS explains what a name currently points at, and a live check explains what the host actually does when you talk to it. This page tells you which network you are dealing with — which, most of the time, is the question you actually had.