Intent data for SaaS: a buyer's guide

Intent data for SaaS is any dataset sold to tell you which companies are researching a purchase right now, and the whole buying decision comes down to one question: when a score fires, can you see why?

Most of the category can't answer that, and the vendors are more candid about it than their resellers are. So this guide is built from the incumbents' own pages, and every number carries its source.

What is intent data for SaaS?

Intent data for SaaS is behavioral data sold to show which companies are researching a purchase. It comes from five structurally different production mechanisms: programmatic ad bidstream, publisher co-ops, owned editorial networks, review sites, and reverse-IP website de-anonymization. Each observes something different, and most of them resolve to an organization rather than a person.

A publisher co-op tags member sites and resolves visitors to companies. Bombora describes its own product this way, on its Our Data page:

By detecting how many users from an organization are researching relevant topics, how frequently they're reading, and how in-depth they're researching compared to their normal activity, Bombora is able to accurately detect when an organization is actively researching a topic.

Read the unit of analysis. It's the organization, and it's counting users inside it. That's a design choice the vendor states plainly, not an accusation from a competitor.

An owned editorial network observes registered members on properties it controls, which is why it can offer something person-level. Informa TechTarget's Priority Engine reports 58 million permissioned members across 220 media properties and sells both prospect-level and account-level intent. A review site observes behavior on its own domain. G2 Buyer Intent asks which of the 100 million buyers researching on G2.com are in market, speaking in accounts throughout.

Reverse-IP vendors resolve visitors against identity graphs, and one of them draws the boundary for us. RB2B states that "RB2B's Person-Level Identity is a US-only technology with a US-only database," attributing the restriction to GDPR and CCPA compliance. A vendor fencing its flagship capability at a border on privacy grounds has told you something about the legal footing of the mechanism.

Bidstream sits on shakier ground still. The Belgian data protection authority fined IAB Europe 250,000 euros on 2 February 2022, finding that the real-time-bidding consent string "can be linked to the IP address of the user, therefore making the author of the preferences identifiable." That's the consent plumbing under programmatic advertising, which is the plumbing under bidstream intent.

What is the difference between first-party, second-party, and third-party intent data?

First-party intent is behavior on properties you own. Second-party is behavior on someone else's property that they sell you directly, like a review site. Third-party is aggregated from networks you have no relationship with. The taxonomy has no standards custodian, so every vendor's definition places its own product in the flattering tier.

That means the useful comparison isn't the tier. It's what the signal resolves to, and whether it explains itself.

	First-party	Second-party	Third-party	Public community
Where it comes from	Your site, product, docs	A review site or publisher selling its own audience data	Bidstream, publisher co-ops, aggregators	Reddit, Hacker News, GitHub, Stack Exchange, Discourse
Resolves to	A session, sometimes a known contact	An account researching a category	An account, usually an IP resolved to a company	A named handle, one person
Can you see why it fired?	Yes, it is your own log	Partly, the category and comparison are named	Rarely, a surge score against a baseline	Yes, the buyer wrote the reason
Evidence you can quote	A page view	A comparison event	An elevated number	The post itself, at a permanent URL
Consent basis	Your own notice and tracking stack	The publisher's terms	Contested, see the IAB Europe ruling	Published in public by the author
Published entry price	Your analytics bill	Quote-based	$200 to $2,750 a month where published, none at 6sense or Pocus	Free to read

The row that decides most purchases is the third one. A first-party log and a public post both explain themselves. A surge score doesn't, and the people selling scores now say so out loud. Pocus CEO Alexa Grabell, writing about why signals alone stopped working, describes reps staring at "Account A at 73, Account B at 68, Account C at 81... but nobody knows what those numbers actually mean."

Where do public community signals fit?

Public community signals sit in a fourth quadrant the priced category doesn't occupy. A forum post is person-level, already public, and carries its own explanation: the buyer states the problem in their own words at a permanent URL. There's no identity-resolution step to get wrong and no surge score to interpret.

The volume isn't marginal, either. SurveyMonkey and Reddit surveyed 1,202 US decision-makers between December 2025 and January 2026 and found 32% of software buyers use Reddit for vendor research. On the developer side, GitHub's Octoverse 2025 reports more than 180 million developers and 5.5 million issues closed in July 2025 alone.

Set that against what identity resolution recovers. Unify, which sells go-to-market orchestration, reports company-level website matching at 30% to 65% of US B2B traffic and person-level at 5% to 20%. Those are vendor estimates with no independent audit behind them, which is how every match rate here should be read. The community conversation isn't under-served by intent tooling. It's off the map.

The market's own packaging complicates the tidy version of this argument. Common Room is a community-signal platform, and it bundles Bombora Company Surge topics into every tier, 6 on Essential and 25 on Enterprise. These aren't opposites. They're different quadrants at very different prices, and one platform sells you both.

When should you not buy intent data for SaaS?

Don't buy it when you sell to small companies, when your team can't work an account without knowing which person to contact, or when nobody will audit the score. Account-level surge data prioritizes a list. It doesn't tell you who, or what to say, or whether the window is open this week.

Coverage. IP-to-company mapping degrades exactly where small and remote-first companies live, and no independent measurement of that gap exists in either direction. If your ICP is 15-person teams on residential connections, ask any vendor for the match rate on a sample of your own target accounts before you sign.

Explainability. 6sense, which sells intent data, published this: "Topic-level intent tells you an account is broadly 'researching CRM software.' That's almost meaningless on its own," adding that "one signal isn't a sales trigger, it's a marketing cue". If the plan is to route a surge score straight to a rep, the vendor has told you what happens next.

Shared supply. Michael McGoldrick of pharosIQ asked it plainly in December 2024: "When everyone has the same 'intent' source are you really gaining any competitive advantage?" Steve Armenti, in MarTech in May 2026, was blunter about bidstream: "The accuracy is lower than co-op or editorial data, the resolution is mostly account-level (IP-resolved to companies), and the privacy footing is shaky." His other line is the one to keep. Your three closest competitors are buying it, too.

Outcome evidence. There's close to none. The one quantified result on G2's Buyer Intent page is a case study, impact.com cutting cost per lead from $120 to $53. One customer, published by the vendor. No independently replicated intent-lift study was reachable while researching this piece, and the largest independent dataset cuts the other way. Ebsta and Pavilion's 2025 GTM Benchmarks, covering 655,000 opportunities across 387 companies, ranks outbound at 1.05x channel efficiency, behind organic inbound at 1.2x and partner referral at 1.3x.

The anti-intent statistics deserve the same scrutiny. The most-quoted pair, 87% of organizations reporting unreliable or inflated intent signals and only 26% of those converting to qualified opportunities, comes from DemandScience's own 2026 report, by way of MarTech's write-up of it. The report itself is no longer reachable, and DemandScience sells intent data.

We're in the same position. LeadSurface has published no conversion-lift study either, because we do not have one. Anyone in this category quoting a lift number is quoting their own marketing.

How do you act on intent data once you have it?

Treat it as a filter, not a trigger. A surge score narrows the list. Something with a name and a quotable sentence is what a person can open a conversation with. Forrester named contact identification inside intent-showing accounts as the top execution challenge teams reported, and said most providers can't solve it.

Brett Kahnke, reporting Forrester's Q1 2023 intent-data survey: "Identifying specific contacts to target within accounts demonstrating intent was cited as the top execution challenge... few intent providers are able to deliver signals from known contacts." Three years on, that gap is where most intent programs stall.

The timing argument underneath it matters more than speed does. 6sense's 2025 B2B Buyer Experience Report, surveying roughly 4,000 buyers at a median purchase size of $200,000 to $300,000, found 95% of winning vendors were already on the buyer's Day One shortlist and 79% of first engagements were buyer-initiated. Their own reading is that being contacted first is a symptom of already being preferred, not the cause of it. The goal isn't to win a race to the inbox. It's to be in the consideration set before it closes, which means catching the moment someone says their current tool stopped working.

That moment predates the category. Craig Elias named it the Window of Dissatisfaction in 2010: after the buyer decides what they have is no longer sufficient, but before they have started doing anything about it. A funding round implies that dissatisfaction might follow. An engineer posting that their observability bill tripled is the dissatisfaction, stated and timestamped, with a person attached.

This is where LeadSurface fits, said once. It watches Reddit, Hacker News, GitHub, Stack Exchange, and Discourse forums, classifies each post into a signal type, scores it, and hands you the ones worth a human reply with a link back to the source thread. The link is the point. Every claim it makes about a lead is checkable in about four seconds.

If you already own intent data, run both for a month. Take your top ten surging accounts and try to name one person to contact at each, then take ten community threads and time the same task. Whichever list produces more real conversations is the one to fund.

A surge score tells you a building is warm. A forum post tells you which person, what broke, and what they tried.