ivorycomBook a demo

Definition

What is visitor intent?

Almost everyone who reaches a B2B website leaves without identifying themselves. Visitor intent is the practice of reading what those sessions imply — which is genuinely useful and routinely oversold, usually in the same sentence.

By Ivorycom7 min read

The Ivorycom visitor view showing identified companies, pages viewed and session recency

The definition, and the resolution it works at

Visitor intent is the inference you draw about a company's buying process from how people from that company behave on your site. The critical qualifier is the level of resolution: in B2B this is a company-level signal, not a person-level one. Knowing a mid-sized logistics firm read your pricing page three times this week is useful. Believing you know which individual did it is usually a claim the underlying data does not support.

  • The unit is the account, not the individual.
  • It is an inference about a process, not an identification of a person.
  • Treating a company-level signal as a person-level one is how teams end up making calls that feel invasive and land badly.

What the signal is actually made of

Four ingredients, and they contribute very unequally. Most of the predictive value sits in the last two, which is the opposite of how the category is usually demonstrated.

  • Firmographic resolution — which company this session probably belongs to, with a confidence attached.
  • Which pages, weighted by what they imply. Pricing, security and integration pages carry meaning that a blog post does not.
  • Depth within a visit. Reading four related pages in one session is a different act from bouncing through four.
  • Repetition across sessions and people. One visit is noise. Three visits from three devices in a week is a signal.

What it can honestly tell you

Three things, reliably enough to change what you do on a Monday morning. None of them is who to call today, which is what it is most often sold as.

  • Timing. An account that was quiet and is now reading is doing something, and the window is usually short.
  • Concern. The pages chosen say what the open question is — price, security, whether it integrates with what they run.
  • Breadth. Several people from one company reading in the same week suggests an evaluation rather than a curiosity.

What it cannot tell you, however good the vendor is

Being explicit about this is what separates a program that survives contact with a sales team from one that gets switched off in month two after a few embarrassing calls.

  • Whether they are actually in a buying process. Competitors, candidates, students and your own staff all read your site.
  • Who to contact. Company-level resolution does not name a person, and the ones that appear to are usually inferring.
  • Budget, authority or timeline. None of that is visible in a page view.
  • Anything at all about a company too small or too remote to resolve. Coverage is uneven and vendors rarely lead with that.

The privacy line, stated plainly

This is not a legal appendix to the topic; it is part of the definition. A visitor intent program that quietly crosses into person-level identification without consent is both a compliance problem and, more immediately, a commercial one — the first call that reveals you were watching an individual costs more than the pipeline it produced.

  • Company-level inference from network and firmographic data is a different activity from identifying an individual.
  • Consent and disclosure obligations apply to the visitor, and they vary by jurisdiction — check rather than assume.
  • Never reference specific browsing behavior in an outbound message. It is the fastest way to make a useful signal unusable.
  • Use it to decide who and when. Use something the buyer told you to decide what to say.

How to act on it without being strange

The most common implementation failure is not technical. It is treating the signal as a script rather than as a prioritization input, which produces outreach that is accurate and unwelcome at the same time.

  • Let it order the queue. It is excellent at deciding who gets attention today.
  • Let it choose the topic, not the opening line. If they read the security page, lead with security — do not say you saw them read it.
  • Combine it with a first-party signal before acting on anything cold. Intent plus a known contact is a different proposition than intent alone.
  • Set a decay window. Intent goes stale in days, and acting on a three-week-old signal is worse than not acting.

What a visitor signal supports, and what it does not

The left column is where these programs earn their keep. The right column is where they generate complaints.

QuestionCan the signal answer it?Why
Which accounts deserve attention this weekYesRelative recency and depth rank well
What topic to lead withUsuallyPage choice implies the open question
Whether an evaluation is under wayOftenMultiple people, multiple sessions, related pages
Which individual to contactNoCompany-level resolution does not name a person
Whether they have budgetNoNot observable from browsing
Whether they are a buyer at allNoCompetitors and candidates read the same pages
What to say in the first lineNo — and do not tryReferencing browsing behavior reads as surveillance

Frequently asked questions

Is visitor intent the same as buyer intent?

Visitor intent is one source of buyer intent. Buyer intent is the broader category covering every signal that a company is in a buying process, including third-party research data, review-site activity and first-party engagement. Visitor intent is the first-party, on-your-own-property subset — which makes it the most reliable and the narrowest.

How accurate is company identification?

It varies enormously by company size, geography and network, and honest vendors publish a confidence level rather than a match. Large organizations on corporate networks resolve well; small companies, remote workers and consumer connections often do not resolve at all. Ask for coverage on accounts that look like your customers rather than an overall match rate.

Does using visitor intent data comply with GDPR?

Company-level inference is a different activity from identifying an individual, but the obligations attach to the processing you actually do and vary by jurisdiction, so this is a question for your own counsel rather than a vendor's marketing page. The practical rule that keeps programs out of trouble is to stay at the account level and never reference an individual's browsing in outreach.

How long does a visitor signal stay useful?

Days, not weeks. Intent decays much faster than fit, which is why a stale signal is worse than none — it produces outreach timed to a moment that has passed. Set an explicit decay window and let records fall out of the queue rather than accumulating.

Look at a week of your own traffic.

The honest test is how many of the companies resolved are ones you would actually want to talk to.