Guide / Buying Signals

Buying intent data, without the hype

Intent data is the most oversold category in the GTM stack. The pitch is seductive: a list of companies that are in-market for what you sell, ready to be called. The reality is messier, more useful in some forms than others, and easy to overpay for. This guide separates what is real from what is vendor noise: the three kinds of intent data, what each is actually good for, and how we use it without letting it drive the bus.

What buying intent data is

Buying intent data is any signal that a company is actively researching a problem or a category. The promise is timing: instead of guessing who might care, you reach the accounts showing behaviour that suggests they care right now. It is a subset of the broader world of signal-based selling, focused specifically on research behaviour rather than on events like hiring or funding.

The category is real and it is useful. It is also wrapped in more marketing than almost anything else you will buy for your stack, so the first job is to understand which kind of intent you are actually being sold.

First, second and third-party intent

The three kinds are not interchangeable. They differ in accuracy, in cost, and in how much of the work you have to do yourself.

• First-party intent. Behaviour on your own properties: website visits, pricing-page views, repeat visits to a feature page, a demo request that stalled. This is the most accurate intent data in existence, because there is no inference. The person came to you. It is also the most under-used, because most teams never wire it into outbound.
• Second-party intent. Behaviour on a property you do not own but have a direct relationship with: a review site where prospects compare you to a competitor, a community, a partner platform. Narrower, but high quality when you can get it.
• Third-party intent. Aggregated behaviour across a network of sites, sold by a vendor who infers that a company is researching a topic. This is what most people mean when they say "intent data", and it is the noisiest of the three. Useful as a prioritisation layer, dangerous as a trigger on its own.

Start with first-party, almost nobody does

The cheapest, most accurate intent data you will ever have is already sitting in your own analytics, unused. Someone visited your pricing page three times this week. Someone from a target account read two feature pages and left. A trial signed up and went quiet. These are not inferences. They are people raising a hand, and most companies do nothing with them because the data lives in a tool that does not talk to outbound.

The first thing we do on the intent side for a new client is almost never to buy a third-party feed. It is to wire first-party signals (website de-anonymisation where it is compliant, product events for SaaS, stalled demo requests) into the same workflow engine that runs the rest of outbound. The hit rate on first-party intent is multiples higher than anything you can buy, because it is the only intent that is not a guess.

The truth about third-party intent

Third-party intent is sold as "these accounts are in-market". What you are actually buying is "somebody at this company, maybe, read something tangentially related to your topic on a site in our network, and our model thinks that means intent". Sometimes that is true. Often it is a researcher, a student, a competitor, or a junior employee with no budget reading a blog post.

This does not make it useless. It makes it a prioritisation layer, not a trigger. Used well, third-party intent re-ranks your existing ICP list so the accounts showing topic activity float to the top of this week's queue. Used badly, it becomes the whole targeting logic, and you end up messaging accounts whose only crime was that an intern read an article. The rule we follow: third-party intent can move an account up the queue, it can never put an account on the queue that fit did not already justify.

How we use intent data in practice

1. Fit first. Build the ICP list the normal way. Intent never overrides fit.
2. First-party as a trigger. Wire your own website and product signals straight into outbound. These can fire a workflow on their own, because they are not a guess.
3. Third-party as a re-ranker. Layer a topic feed on top of the fit list to decide what gets attention this week. Boost, do not gate.
4. Verify before you fire. Before a message goes out on an intent signal, a quick check confirms the account is real and current. Intent feeds go stale fast.

We run this in usegrit.io, where first-party signals, third-party feeds and the rest of the enrichment live in one place and feed one scoring model. Clay handles the same job if that is your stack. The platform is not the point. The discipline is: fit gates, first-party triggers, third-party re-ranks, and everything gets verified before it sends.

Before you buy a third-party feed

Three questions save most of the money teams waste here. What is the data source, specifically, and can the vendor explain how a topic maps to a buying decision in your category? Can you trial it on a sample of accounts you can verify by hand, before signing? And do you already use your first-party intent, because if you do not, you are buying a worse version of data you are ignoring for free.

Most teams should answer the third question before they ask the first two. For where intent sits in the wider approach, see signal-based selling. For how it plugs into a full build, the how to build a GTM system guide has the seven-step frame.