Intent data tells you a company researched a topic somewhere on the web. It does not tell you who researched it, why, or whether they will ever buy. Signal data is the other half: an observable event at the account, a new hire, a price change, a tool swap, that you can see and date.
What is intent data?
Intent data is evidence that an account has been reading about a problem you solve.
ZoomInfo defines it as behavioral data that shows when a company, or a person at that company, is actively researching a topic tied to a business problem, product or solution.
One word in that definition carries the weight: topic. The unit of measurement is a subject area, not a product and not a budget. An account reading about data governance has not told you it is changing vendors this quarter.
It splits three ways. First-party comes from channels you own or pay for, including your website, emails, events, product and ads. Second-party is another company's first-party data, shared through a direct partnership. Third-party is collected across external websites and content networks, and that is what people buy.
How is B2B intent data collected?
Third-party intent runs on pooled web behaviour, and the pools overlap more than the branding suggests.
ZoomInfo names three mechanisms. Data cooperatives, where a network of publishers and B2B sites pools engagement data in a consent-based model. Publisher direct, where a single publisher, or a group of niche sites, shares data on its own audience. And bidstream, data captured during real-time ad bidding, which its own page says carries more noise.
Bombora says its Data Co-op reflects real research across 14B+ monthly events and 5M company domains, and that 86% of the data in our Co-op is shared exclusively with Bombora.
Then work out whose pool you are actually buying. Demandbase sells its own intent product and claims direct access to the bidstream, the source for the most intent signals. The same page offers to Import Bombora Surge automatically every time there's an update, Import G2 intent and Import TrustRadius intent. Ask every shortlisted provider which part of its feed it collects and which part it rents from a pool your competitors can rent too.
Identification is the other soft spot. Influ2's round-up says 6sense uses IP/Cookie data as the identification method, which it calls not always particularly accurate. Influ2 sells a competing product and lists itself first, so read that as a rival's criticism rather than a test result.
How is signal data different from intent data?
Signal data is an event, not an inference.
Signal-based selling means acting on evidence instead of a schedule: you contact an account because something happened there. The events usually named are funding rounds, leadership changes, relevant hiring activity, product launches, regulatory decisions. Each has a date, a source page and someone who owns it internally.
| Intent data | Signal data | |
|---|---|---|
| What it is | A topic an account appeared to research | An event that happened at the account |
| Where it comes from | A co-op, publisher direct feed or bidstream | Job boards, public filings, company pages, your CRM |
| Resolution | Account level, often inferred from IP or cookie | A named company, with a date and a source |
| What it is good for | Choosing which accounts to work this month | Naming the reason you are calling today |
| Where it fails | Generic topics, or an ICP too broad to act on | Quiet accounts, where nothing has changed to see |
The two are not substitutes. Intent is a prioritisation input. A signal is a reason to open the email with something other than your product.
How accurate is third-party intent data?
Nobody publishes an average-case number, including the companies selling it.
The category publishes its best case instead, often with nothing behind it. Factors.ai's platform round-up asserts that companies using intent data see conversion rates triple and cites no study, sample or method for it. The Starr Conspiracy, a B2B marketing agency whose page states that we don't sell intent data, names the problem: every vendor-published intent statistic has a selection bias problem, because the vendors publishing them are measuring cases where their product worked, and their published match rates are best-case, not average-case.
Sentiment is easier to source than precision. DemandScience surveyed over 250 demand generation and B2B marketing professionals and reports that 48% of Marketers affirmed that intent data is a critical part of their sales and marketing strategy while 94% of B2B Marketing Leaders are Overwhelmed and Exhausted from Trying to Make Intent Data Work. DemandScience sells in this category, so read it as the industry describing itself.
Even the vendors concede the limit. Demandbase, which sells an intent product, published this on its own blog: a single search on a term does not mean an account is ready to buy (or even more likely to buy).
Should a B2B outbound team pay for intent data?
It depends on whether you have more accounts than you can work.
Intent data is a sorting tool, so the first test is whether you have anything to sort. It earns its cost when your named market is large enough that choosing badly is expensive, and your ICP tight enough that a topic spike inside it means something. We use roughly 1,000 named accounts as our own cutoff for whether a market needs sorting at all. That is our number, not an industry benchmark: below it you can work the whole list, so the money goes further on research per account than on ranking them. If your ICP is a sector rather than a segment, a spike tells you a stranger read an article.
At the enterprise end, pricing is quote-only. Bombora's pricing page carries no figure, plan ladder or starting price, and routes you to speak to an expert instead. Budget for a quoted annual contract and for the work of acting on the feed, the line item most teams forget.
The other failure case is organisational. Without someone accountable for turning a surge into outreach within days, the subscription is a report.
Can you detect buying signals without buying intent data?
For the observable half, yes.
Hiring, funding, pricing changes, leadership moves and tool adoption are public. Access is not the hard part. Maintenance is: watching a defined account list, noticing what changed, and keeping the record current.
The base rate is published, if only by a vendor. Salesmotion, which sells signal tooling, analysed 8,518 companies and found 85% produced at least one signal over the period it measured, while 30.8% generated either a hard trigger event or relevant hiring activity, with a median gap between successive trigger events at active companies of 10.5 days. Treat the levels as directional. The shape is the useful part: most accounts do something, and under a third do something you would call about.
Tracking that by hand works until one person can no longer re-check the whole list.
Pulse is ours, so treat this as disclosure. It researches approximately 100 attributes per company, each with its source and a confidence level, and reports hiring, repositioning, pricing and tooling changes as dated signals, scored from 0 to 1 for how well the evidence matches, a score its own page says is to rank signals, not to predict a deal. The difference from a bought feed is the loop: a co-op sells the same pooled behaviour to everyone inside it and never learns whether it worked, while your own replies and meetings tell you which signals were worth acting on.
If the open question is which accounts to work, choosing a sales intelligence platform covers the wider category. If it is the contract you already signed, what Apollo and Clay actually replace is the more useful page.
Frequently asked questions
Who are the main intent data providers?
Influ2's round-up names ten: Influ2, Pendo, Snitcher, G2 Buyer Intent, Gartner Digital Markets, 6sense, Demandbase, ZoomInfo, Bombora and Cognism. It is published by a vendor that lists itself first, which is normal for this keyword. Before shortlisting, check which of them run their own collection and which resell or import a co-op, because several front the same upstream pool.
How does ZoomInfo determine intent?
Its own page describes a relative measure rather than an absolute one: ZoomInfo measures each account's activity against its own history and flags accounts whose interest is rising, weighing the amount of content consumed, the number of people researching, the type of content, time on page and how recently the activity happened. A surge means busier than usual for that account, not busier than other accounts.
How long is an intent signal still worth acting on?
Not long. The Starr Conspiracy reports that third-party intent signals go stale quickly, that many operators plan against a 7 to 14 day working window, and that a common surge-to-touch time (target: under 48 hours) is the bar teams set. If your routing takes a week, you are calling about research that has already finished.
What is the difference between intent data and account scoring?
Intent data is one input; account scoring is the model that weighs it against everything else. A score blends firmographic fit, your own engagement history and intent into a single ranking. The practical consequence is that a high score can come from fit alone, with no recent research behind it, so ask which component moved before you treat a score as a reason to call.
How do you tell whether intent data is working for your team?
Measure it yourself, because the published figures will not transfer. Tag the accounts you contacted because of a surge, hold back a matched sample you contact on fit alone, and compare replies and meetings over a quarter. The Starr Conspiracy's advice on this is blunt: if you can't find a hard number you trust, measure your own.