What is website visitor identification?
Website visitor identification matches anonymous visits to a known company or person, so a sales team can see which accounts are researching it before anyone fills out a form. It works at two levels. Company-level identification names the organization behind a visit. Person-level identification tries to name the individual.
The difference matters more than most product pages admit. Company-level identification is an established technique with known failure modes. Person-level identification is a probabilistic match against third-party data, and how often it is right depends heavily on where your visitors are and how they connect.
Most of your traffic never identifies itself. Visitors read the pricing page, compare you with two competitors and leave. Identification gives you a list of the companies behind some of those visits and, for some of them, a guess at who.
How to identify anonymous website visitors: four methods
Every visitor identification tool combines some of the same four methods. They differ in what they can name and how sure they are.
| Method | What it can name | How sure | Where it fails |
|---|---|---|---|
| Reverse IP lookup | The company that owns the network | Good on office networks | Home internet, VPNs, mobile, shared offices |
| First-party cookies | The same browser coming back | High, until cookies are cleared or blocked | Browser privacy settings, new devices, consent banners |
| Identity graphs | A likely person | Varies widely, mostly US traffic | Outside the US, shared devices, stale data |
| Forms, logins, email clicks | The actual person | Certain | Only covers people who already engaged |
Reverse IP lookup
Every visit comes from an IP address. Companies that run their own networks own ranges of addresses, and databases map those ranges to company names. A visitor identification script reads the address, looks it up and returns the company, often with its industry and size.
This is the core of B2B visitor identification. It needs no data from the visitor beyond the connection itself, and on corporate networks it is usually right. It is wrong, or silent, whenever the address does not belong to the visitor's employer. That covers anyone working from home on a consumer broadband line, anyone on a VPN, anyone on a phone network, and anyone in a coworking space or a building with shared internet.
First-party cookies
A first-party cookie is a small file your own site sets in the visitor's browser. It does not tell you who someone is. It tells you that this browser has been here before, so visits over several weeks can be tied into one history: which pages, how often, in what order.
Cookies make the other methods more useful. When a returning browser later fills out a form, the earlier anonymous visits can be attached to the new contact. Browsers such as Safari limit how long some cookies last, people clear them, and consent banners in the EU block them until the visitor agrees. Some tools add browser fingerprinting, which recognizes a device from its configuration without a cookie. In the EU that needs consent too (see the privacy section).
Identity graphs and person-level matching
Person-level tools go further. They match the visitor against an identity graph: a third-party data set that links devices, browsers, IP addresses and hashed email addresses to people, built from sources such as publisher logins, newsletter sign-ups and data partnerships. When a visit matches a device the graph has seen before, the tool returns a name and often a LinkedIn profile.
This is where the big match-rate claims come from, and where most of the errors live. The graphs are strongest for US consumer devices, which is why person-level products mostly limit themselves to US traffic. A match can be stale, because people change jobs, or wrong, because devices are shared. Treat a person-level result as a hypothesis about who it might be, and check it before you act.
Forms, logins and email clicks
The only certain identification is the one the visitor gives you. A form fill, a login to your product or a click on a tracked link in an email you sent ties the browser to a known contact. With first-party cookies in place, their earlier anonymous visits come with them.
This method only covers people who have already engaged. That makes it the baseline you can test the other methods against.
What visitor identification gets wrong
A tool that names a company or a person does not tell you how sure it is. These are the ways the results go wrong, and they are worth knowing before you route a single account to a rep.
Home and VPN traffic. Remote and hybrid work moved a large share of B2B browsing off company networks. A visit from a home broadband line resolves to an internet provider or to nothing, and a visit through a corporate VPN can resolve to the VPN provider's data center or to the wrong office.
Shared and cloud networks. Coworking spaces, office buildings, universities and cloud hosting all put many organizations behind the same addresses. A tool that maps an address to "the company in suite 400" can be wrong for everyone else in the building. Traffic from cloud providers is often bots, not buyers.
Person-level guesses presented as facts. A name on a screen looks certain. A person-level match is a probability that the tool has rounded up to an answer. The cost of a wrong name is high: a rep emails someone who never visited, about pages they never read.
Match rates that count the wrong thing. Vendors quote match rates as the share of visits they returned something for. That number says nothing about whether the matches were right, or whether the matched companies are ones you would ever sell to. A tool that identifies 40 percent of your traffic, mostly students and job seekers, is worth less than one that identifies 15 percent and gets your target accounts right.
One visit read as intent. One page view from a large company tells you almost nothing. Someone in a 20,000-person company reads your blog every day for reasons that have nothing to do with buying. Repeat visits, several people from the same account, and visits to pricing and comparison pages carry the signal. A single visit does not.
How to test a visitor identification tool before you trust it
Most tools offer a free tier or a trial. Use it to measure the tool against visitors you already know, on your own site, before you believe anyone's match rate.
- Install the script on your site for two to four weeks of normal traffic.
- Pull the visits you can already identify: form fills, product logins and clicks from your own emails. These are your known answers.
- For each known visit, check what the tool said before the visitor identified themselves. Count how often the company was right, and separately how often the person was right.
- Look at the matches you cannot verify. What share are companies that fit your ideal customer profile, and what share are internet providers, universities, competitors and your own staff?
- Divide what the tool costs by the number of correct, in-profile accounts it found. That is the number to compare between tools.
Step 3 is the one no vendor will run for you. It turns a headline match rate into "on our site it named the right company two times in three and the right person one time in five", which is a number you can plan around. The figures in that sentence are an example of the output, not a benchmark.
Is identifying website visitors legal?
Identifying company visitors is a common B2B practice, but the rules depend on where your visitors are, and person-level identification raises the stakes. This section is a map of the issues, not legal advice.
In the EU and UK, an IP address can be personal data. The Court of Justice of the EU held in Breyer (case C-582/14, October 2016) that even a dynamic IP address can be personal data when the site operator has legal means to link it to a person. Separately, the ePrivacy rules require consent before storing or reading information on a visitor's device. The European Data Protection Board's Guidelines 2/2023 on the technical scope of that rule (version 2, adopted October 2024) state that it covers tracking pixels, tracking links, fingerprinting and some IP-based tracking, not only cookies. For EU traffic, that means a consent banner that actually gates the tracking script, a privacy notice that says you identify visiting companies, and a lawful basis under GDPR for any outreach that follows.
In the US, California's privacy law (CCPA, as amended by CPRA) treats IP addresses and browsing history as personal information. Visitors have the right to know what you collect and to opt out of its sale or sharing, and businesses must honor opt-out signals such as Global Privacy Control. Other states have similar laws. Email outreach also has to follow CAN-SPAM: accurate sender details, a working unsubscribe, and a physical address.
Whatever the jurisdiction, the practical rules are the same. Say in your privacy notice that you use visitor identification. Respect consent and opt-out signals. Keep a suppression list and use it.
What to do with an identified visit
An identified visit is the start of research, not a reason to call. Four steps turn it into pipeline.
Check fit first. Score the company against your ideal customer profile before anyone spends time on it. Most identified companies are not buyers, and the ICP is what lets a team ignore them quickly.
Weight the pages and the pattern. Pricing, comparison and integration pages carry more intent than blog posts. Several visits in a week, or several people from the same account, carry more than one visit. Fit tells you who is worth the effort. These patterns help tell you when, alongside triggers and research signals, because fit and timing are separate questions.
Route it to one owner. An identified account with no owner is a dashboard entry. Decide in advance who gets which accounts, by territory, segment or a round robin, so the follow-up happens while the visits are recent.
Reach out about the problem, never the visit. Do not open with "I saw you on our pricing page". In Thomas's original version of this article, he described a composite of client cases: teams that opened their emails by citing the visit got poor response rates, and some prospects replied asking whether they were being spied on. Write to the person in the right role about the problem the pages they read address. If they want to talk, the visit was the reason you were on time. They do not need to know it.
If you want to compare vendors, CIENCE keeps a review of website visitor tracking tools. graph8 Visitor Identification is free in graph8: it matches visits to companies by IP and, where it can, the person, shows the pages each account viewed, and routes the account to an owner with Slack alerts and a CRM log.
Name the company, check the fit, read the pattern, and write about the problem. The visit is why you are on time, not what you talk about.