Intent data is a set of signals suggesting that a company is researching a problem or a category right now: content consumption, search behavior, review site activity or actions on your own properties. First party signals are evidence. Third party signals are probability, aggregated to the account and frequently to a whole building.
What it is
Three tiers, with very different reliability. First party intent is what happens on your own site and in your own funnel: a pricing page visit, a demo request, a reply. Second party comes from a partner such as a review marketplace, where the buyer's action was real but happened elsewhere. Third party is modeled from publisher networks and resolved to a company by IP or by identity graph, which is why the signal points at an organization rather than a person.
How it works
Providers watch content consumption across a network of sites, group it into topics, and report when an account's consumption of a topic rises above its own baseline. That baseline is the whole trick: the interesting output is a surge, not a level. The resolution step is the weak link. A signal attributed to a company may come from a contractor, a job applicant, a competitor or an intern writing a paper, and the platform cannot tell you which.
Why it matters for winning clients
Timing is the variable outbound teams control least. A message that lands during an active evaluation converts at a different rate than the same message sent at random, which is what makes intent commercially attractive. Used well it reorders a queue you already trust. Used badly it becomes a subscription that manufactures urgency about accounts that were never in your market.
Example in LeadCanvas
A small agency without an intent subscription uses observable proxies instead: a new location on a listing, a burst of recent reviews, a job posting for a role that implies the problem it solves, a site that has visibly not changed in years. These are weaker than modeled intent and far easier to verify, because each one is a fact on a public page rather than a probability delivered in a report.
Common mistakes
Opening a message with the signal itself, which reads as surveillance and kills the reply. Buying intent before the ideal customer profile is settled, so the platform prioritizes accounts you cannot serve. Treating a surge as an account level fact when it may be one anonymous visitor. And skipping the counterfactual: if your close rate on intent flagged accounts matches your close rate on the rest, you are paying for a reordering that does nothing.
Related terms
Lead scoring is where an intent signal should land, as one input among several. A sales trigger event is the observable cousin of modeled intent. The ideal customer profile is the filter that has to run first. The links below open those entries.