Intent data explained, including what it cannot tell you

Intent data is the most oversold category in the go to market stack, which is unfortunate, because used narrowly it genuinely works. The confusion comes from vendors describing very different things using the same word.

The main types, and what each actually observes

Topic or content intent

Publisher networks track which organisations read content on a given subject, usually resolved from IP address to company. A spike relative to that account's baseline is reported as a surge. It tells you someone at the company read something. It does not tell you who, or whether they have any budget.

Review and comparison intent

Activity on software review and comparison sites. This is a later stage signal and generally the highest quality of the three, because someone comparing vendors is further along than someone reading an article. Volume is correspondingly much smaller.

Event and trigger signals

Not inference at all, but observable facts: a funding round, an executive hire, a technology change, a job posting naming a tool. These are the most defensible signals because they are verifiable, and they often predict budget more reliably than content consumption.

The limits worth knowing before you budget

Three constraints apply to nearly all intent data:

• It is account level, not person level. An account showing intent still requires you to identify who owns the problem. Intent narrows the list; it does not hand you a contact.
• IP resolution is imperfect. Remote work, VPNs, and shared networks all degrade the company match. Accuracy varies significantly by provider and by region.
• It is probabilistic. A surge means the account is more likely than baseline to be in market. It is not a confirmation, and treating it as one produces badly calibrated outreach.

Where it earns its cost

Intent is a prioritisation layer, not a list source. Its value is in ordering an existing target account list so limited rep capacity is spent on the accounts most likely to respond. Used that way, on a defined ICP, it reliably lifts connect and reply rates.

It performs poorly as a prospecting source. Buying intent against a broad, undefined market produces a long list of accounts that read something once, and working that list burns rep time for little return.

Stacking signals raises precision

A single surge is weak. Several independent signals on the same account within a short window are considerably stronger: a funding round, plus hiring for roles that use your category, plus review site activity. Combining triggers narrows the list, which is usually what you want, since capacity is the constraint.

Buying it sensibly

Ask any provider three questions before signing. Which specific signals underlie the score, since a composite number hides whether it is content, review activity, or events. How is the account resolved from raw traffic, and what accuracy do they claim by region. And what is the refresh cycle, because a signal surfaced three weeks late is a signal your competitor already acted on.

Because intent involves observing behaviour, sourcing and consent documentation matter. The GDPR is the relevant framework for European activity, and providers should be able to explain their lawful basis.

Common questions