The short answer
An IP address identifies the network a visit came through, not the person. If that network belongs to a company, a lookup can name the company. If it belongs to a broadband provider, a mobile operator, a VPN or a cloud host, the lookup can only name the provider, and a good tool reports nothing. That is why vendors report company-level match rates anywhere from about 20% to 65% depending on the traffic⁶, and why the same tool performs differently on different websites.
This page explains the mechanics behind website visitor identification. For the practical set-up, see how to see which companies visit your website.
Where IP addresses come from
Every device that talks to your website does so from an IP address, and every public address is part of a block handed down through a chain of registries.
- IANA manages the global address space and allocates large blocks to the five regional internet registries. It does not normally allocate addresses to internet providers or end users directly¹.
- Regional internet registries allocate blocks to their members. For the UK and the rest of Europe this is the RIPE NCC, which keeps a public record of the number resources registered in its region².
- Local internet registries, usually internet providers and large organisations, assign addresses from their blocks to their own networks and customers.
- End networks, such as an office, a home or a mobile phone, use those addresses to reach the internet.
The RIPE NCC ran out of new IPv4 addresses to allocate on 25 November 2019³. Since then IPv4 space has been scarce, transferred and reused, and providers share single addresses among many customers. Both make the link between an address and a single organisation weaker than it was.
What a lookup checks
A commercial IP-to-company service combines several kinds of evidence. None of them is complete on its own.
1. Registry records
The RIPE Database holds the allocations and assignments of address space in its region, each with a network name and, often, a link to the holding organisation⁴. When a large company, university or public body holds its own block, the record names it, and the match is straightforward. When a provider has assigned a small block to a business customer, the record may name the provider only, or carry a network name that hints at the customer. The RIPE NCC notes that some details, such as the country attribute, should not be relied on to be fully accurate⁴.
2. Reverse DNS
Many addresses have a reverse DNS (PTR) record, a hostname the network operator has attached to the address. Corporate networks sometimes use their own domain, which points to the company. Broadband providers usually use generic hostnames containing their own name, which points nowhere useful.
3. Vendor mappings
This is where vendors differ most. Each builds a proprietary map of addresses to companies from sources such as registry data, its own customers' traffic, form fills and logins associated with addresses, and commercial data partners. HubSpot puts it plainly: every IP tracking tool maintains its own proprietary list of companies and IP addresses, so the lists differ between tools⁵. The map has to be refreshed constantly, because companies move offices, change providers and move to cloud services.
4. Classification
Before reporting a company, the service classifies the address: business network, residential broadband, mobile network, hosting or cloud, VPN or proxy, or known bot. Only the first category should produce a company name. Everything else is discarded or shown as the provider, which most tools hide by default⁵.
5. Confidence
Finally, the service decides whether its evidence is strong enough to report. This is a trade-off every vendor tunes. A cautious threshold reports fewer companies with fewer mistakes. A loose threshold reports more companies and more wrong ones.
Why match rates vary
Two websites using the same tool can see very different results, because the visitors are different. These are the factors that move the number most.
Residential connections and remote work
A person working from home connects through their home broadband, and the address belongs to the provider. Leadinfo, a European vendor, notes that consumer providers often assign dynamic addresses, and that home workers are hard to trace unless they connect through a company VPN or fixed address⁷. In Great Britain, 28% of working adults were hybrid working between January and March 2025⁸. Every day those people research from home is a day their visits are likely to stay anonymous.
Mobile networks
Mobile operators put many subscribers behind shared addresses, so a phone on 4G or 5G almost never resolves to an employer. HubSpot states that visits from a mobile network cannot be identified⁵. One vendor puts company-level match rates for mobile traffic at 15% to 30%⁶. StatCounter's 2026 data shows just over half of UK web page views coming from mobile devices⁹, although B2B research leans towards desktop.
VPNs and privacy relays
Consumer VPNs and privacy relays replace the visitor's address with one belonging to the service. Apple's iCloud Private Relay, for example, passes Safari traffic through two relays and gives the website a temporary address, hiding the user's IP address from the sites they visit¹⁰. A corporate VPN is different: if it sends traffic out through the company's own addresses, it can make a remote worker identifiable.
ISP-registered ranges
Most small and medium firms rent business broadband or a leased line, and their address is registered to the provider. HubSpot notes that smaller companies and individuals may not have their own IP address and will be identified as their internet service provider⁵. A vendor can still match these if it has other evidence, such as earlier form fills from that address, but coverage is patchy. For a site that sells to small firms, this is often the biggest single limit.
Company-owned ranges
Large enterprises, universities and public bodies often hold address blocks in their own name or route all staff traffic through their own gateways. These resolve well. One vendor reports that enterprise traffic matches at roughly twice the rate of SMB traffic⁶. A site whose audience is mostly large organisations will see higher match rates than one selling to small firms, with the same tool.
Geography, sector and data age
Vendors usually have the best data where most of their customers are. One US-based vendor reports 20% to 40% at company level for international traffic, against 30% to 65% across its mostly US customers⁶; Leadinfo claims 35% to 40% in Europe⁷. Sector matters too: sectors where staff work mainly from offices match better than those where people research from home. And because address ownership changes, a mapping that is months out of date loses accuracy.
Match rate is not accuracy
A vendor's headline number usually measures coverage: how many visits it labelled. It does not tell you how many of those labels are right. A 2026 vendor benchmark describes the trade-off as tiers, from high-precision tools matching 30% to 45% of sessions with 90% or more precision, to high-recall tools matching 50% to 70% with 60% to 80% precision¹¹. The figures are the vendor's own, with no published method, but the principle is sound: ask for both numbers, and measure both yourself.
| Term | What it measures | What to ask |
|---|---|---|
| Match rate (or fill rate) | Share of visits or visitors given a company name | Per visit or per visitor? Over which traffic? |
| Accuracy (or precision) | Share of named companies that are correct | How was it checked, and on whose traffic? |
| Coverage by region | Match rate for a given country | What is it for UK visitors specifically? |
How to test a lookup on your own traffic
- Seed known visits. Ask colleagues, friendly customers and partners to visit a specific page from office networks, home broadband and mobile, and note which are named correctly.
- Sample the matches. Take twenty identified companies at random and check whether each is plausible for your audience and the pages viewed.
- Count the provider share. Ask how many visits were classed as internet providers. A high share is a sign your audience works from home or from small firms, not necessarily a weak tool.
- Compare two tools on the same days, so differences come from the data, not the traffic.
A note on personal data
The ICO lists IP addresses among the online identifiers that may be personal data under the UK GDPR¹². A lookup that returns only a company still processes the address to get there, so it needs a lawful basis and a mention in your privacy notice, and any cookies the tool sets fall under PECR. Our visitor identification guide summarises the points to consider. This is not legal advice.
Where Artiweave fits
Artiweave continuously improves the website a B2B company already has, with every change approved by a marketer and measured honestly, including when the result is inconclusive. Company-level visitor identification is coming to Artiweave. It is not live yet, and we will publish our own match rates only once we can measure them on real traffic.
Frequently asked questions
Is reverse IP lookup the same as IP-to-company lookup?
Not quite. A reverse DNS lookup returns the hostname attached to an address. IP-to-company lookup is broader: it combines registry records, reverse DNS and a vendor's own mappings to decide which organisation, if any, is using the address.
Why does my visitor identification tool show internet providers?
Because many visitors connect through addresses registered to their provider rather than their employer: home broadband, mobile networks and small business connections. Without other evidence the tool can only name the provider, and most tools hide those visits by default.
Does IPv6 make identification easier?
It can, because IPv6 is plentiful and organisations are less likely to share addresses. In practice it depends on how each network is set up and how well a vendor maps IPv6 ranges, so test it rather than assume it.
Can an IP address identify a person?
An address on its own identifies a network connection, not a person. Combined with other data it can relate to an individual, which is why the ICO treats IP addresses as potentially personal data.
Sources
- IANA: Number Resources
- RIPE NCC: Regional Internet Registry
- RIPE NCC: The RIPE NCC has run out of IPv4 addresses (25 November 2019)
- RIPE NCC: RIPE Database documentation, FAQ
- HubSpot Knowledge Base: Prospects frequently asked questions (updated December 2025)
- Warmly: Website Visitor Identification Match Rates (March 2026). Vendor's own production data.
- Leadinfo: How Reliable is IP Tracking for Identifying Website Visitors? (November 2025). Vendor claim, no method stated.
- Office for National Statistics: Who has access to hybrid work in Great Britain? (June 2025)
- StatCounter Global Stats: Desktop vs Mobile Market Share, United Kingdom (2026)
- Apple Support: About iCloud Private Relay
- Happierleads: Reverse IP lookup accuracy benchmark (June 2026). Vendor's own figures, method not published.
- ICO: What are identifiers and related factors?
Evidence first, for every change.
Artiweave continuously improves the website your B2B company already has, with every change approved by you and measured honestly. Company-level visitor identification is coming. It is in private beta.