What is intent data?

The idea is that research leaves footprints. When several people at one company start reading about, say, warehouse robotics, that company is more likely to be evaluating a purchase than one that never touches the subject. Intent signals try to capture those footprints and turn them into a ranked list of accounts showing unusual interest in specific topics.

First-party and external intent

There are two broad sources. Your own website and content give you first-party intent: who visited pricing pages, downloaded a guide or watched a demo. External intent comes from networks of publishers and sites that share reading activity with a signal provider.

Bombora, a well-known example, describes a data co-op of thousands of media sites and B2B brands, and scores research by comparing an account's most recent three weeks of activity against its own 12-week baseline.

• Account-level lists: company names or domains with topics where research rose.
• Scores or tiers showing how strong the change is compared with normal levels.
• Sometimes role-level hints, such as whether research came from engineering or finance.
• Rarely, if ever, a named individual; most providers resolve activity to a company, not a person.

The scenario below is invented. A company runs a handful of trade publications for the cold-chain logistics industry: news sites, newsletters and a webinar series.

Its analytics show which organisations read which articles, worked out from corporate network ranges and newsletter sign-ups by work domain. A sales-intelligence firm wants to add that readership to its signals, because few other sources cover such a narrow industry.

The broker starts with the reader side. The publisher's privacy notice and cookie banner have to cover sharing reading activity with partners; if they don't, they need updating before any feed starts, and earlier records stay out.

Next comes granularity. The publisher agrees to share topic-level counts per company domain each week, never article-by-article trails of named subscribers. Companies with very few readers are dropped so no individual can be inferred.

The licence restricts resale of the raw feed and requires the buyer to label the signal's source in its product. The publisher gains a steady new line of business from logs it already kept. The buyer gains niche coverage. Neither side pretends the signal proves anyone is about to buy; it is evidence of attention, priced and described as such.

Attention isn't intention. Students, journalists, competitors and existing customers all read about a topic without planning to buy. Remote work makes mapping traffic to companies harder, since home connections don't reveal an employer.

Small companies produce too little activity to register at all. And a spike only says something changed; it doesn't say what or why. Teams that treat a score as a qualified lead tend to waste outreach and annoy prospects.

The privacy side

On the privacy side, the logs underlying these signals are usually about individuals' reading habits before they are aggregated. Browser cookie behaviour matters too: Google said in 2025 that Chrome would keep its existing user choice for third-party cookies, so collection methods depend on browser settings and on consent banners where those are required.