B2B buying signals: which ones actually predict a deal

Every intent data vendor will show you a dashboard where accounts light up. The accounts look like they are in-market. Some of them are.

The problem is that the dashboard does not distinguish between an account whose VP of Engineering just posted three roles that only make sense if they are rebuilding a system you replace, and an account where a summer intern read two blog posts about your category.

Both surface as intent. Only one of them is worth a Tier 1 asset.

This is a ranking of twelve common B2B buying signals by how reliably they predict an actual deal, based on what holds up in practice rather than what sells well in a demo.

What a buying signal actually is

A buying signal is an observable event that raises the probability an account will enter a purchase process in the near term.

The definition has two load-bearing words that most vendor material quietly drops.

Observable. You either saw it happen or you did not. Inferred, modelled and probabilistic signals are a different and much weaker category.

Near term. A signal that tells you an account might buy something in this category eventually is not a signal. It is a description of your ICP.

Hold those two words firmly and most of the intent market thins out considerably.

The two families

First-party signals happen on property you control. A pricing page visit, docs read, a demo started and abandoned, an email reply, a webinar attended, a support ticket, a trial signup that went cold.

Third-party signals happen everywhere else. Job postings, funding, executive changes, technology adoption, review site activity, conference attendance, published content, layoffs.

The consistent finding across teams that measure this honestly: first-party signals are dramatically more predictive than third-party signals, and dramatically fewer teams instrument them properly.

This is a strange market failure. First-party data is free, accurate and yours. Third-party data is expensive, fuzzy and rented. Yet most teams buy the second before instrumenting the first, because the second arrives as a dashboard and the first requires work.

The reliability ranking

Ranked from most to least predictive. The reasoning matters more than the position.

Tier A: act on these immediately

1. Repeat pricing page visits from multiple people at the same account

The strongest signal in B2B, and it is free. One person on your pricing page is curiosity. Three people from the same domain in ten days is a buying committee doing a comparison. Nothing you can purchase comes close to this.

2. An inbound reply that asks a specific implementation question

"How does this handle SSO with Okta" is not a question you ask about a product you are browsing. Specificity is the tell. Generic interest is weak, specific friction is strong.

3. Job postings that only make sense if they are solving your problem

Not "they are hiring, so they have budget". That is firmographic noise dressed as intent. The signal is a role whose responsibilities describe the problem your product removes. A company hiring three people to manually reconcile something your product automates is telling you their current approach is failing, in public, with a date on it.

This is the most under-used third-party signal in B2B, largely because it requires reading rather than a surge score.

4. Technology removed from the stack

Adoption of a competitor is a weak signal. *Removal* of a competitor is a strong one. It means a contract ended or a decision was reversed, and there is a gap right now.

Tier B: real, but needs a second signal to confirm

5. Executive change in the function that owns your problem

A new VP in the relevant seat typically re-evaluates tooling within their first two quarters. The signal is real but the timing is loose, and the new exec is drowning in their first month. Reliable, slow, and worth a patient sequence rather than an immediate push.

6. Funding rounds

Widely over-weighted. A funding round tells you an account has money, not that they have your problem. It raises the ceiling on deal size and tells you nothing about timing. Useful as a *multiplier* on an account that already fired a real signal. Useless on its own, which is exactly how most teams use it.

7. Docs, changelog or integration page visits

Deeper in the product than a marketing page, which makes it more serious than general browsing. But it also catches developers evaluating for a project that never gets funded. Confirm with a second signal.

8. Comparison or alternatives page visits

Strong intent, weak position. Someone at the account is comparing you against a competitor. That is real, but by the time you see it you are usually already in a process that someone else started and framed. Act fast and expect to be catching up.

Tier C: weak. Do not build a program on these

9. Third-party surge scores

The core product of the intent data industry. An account is reading more content about your topic across a publisher network than its own baseline.

The problems are structural, not fixable by a better vendor. The reader is anonymous, so you do not know if it was a decision-maker or an analyst or a competitor. The topic mapping is coarse, so adjacent topics blur together. The baseline is noisy at smaller accounts. And by the time a surge registers, the research has often been going for weeks.

Surge data has real value as a *prioritisation tiebreaker* between two otherwise equal accounts. It has very little value as a trigger. Most teams use it as a trigger.

10. Content downloads and webinar registrations

Measures willingness to trade an email address for a PDF. That is a real behaviour, but it correlates with being a consultant, a student, a competitor or a job seeker roughly as often as with being a buyer.

11. Social engagement

A like on a LinkedIn post is not a buying signal. It is occasionally a *reachability* signal, which is a different and genuinely useful thing. Treat it as a warm door, not a hot lead.

12. Firmographic fit alone

The most common thing mislabelled as intent. "Series B SaaS company with 200 employees in fintech" is a description of your ICP. It contains no information about timing. If your intent dashboard is mostly surfacing accounts that match your filters, you have bought an expensive list.

Combination beats any single signal

The practical lesson: single signals are weak, combinations are strong, and the multiplication is not linear.

A job posting alone is a maybe. A job posting plus two pricing page visits from that domain in the same fortnight is close to a certainty that something is happening.

A useful scoring approach:

Recency decay is the piece most teams omit, and the omission is expensive. A pricing page visit from four months ago is not a diluted version of a fresh one. It is a different thing entirely, usually meaning they already looked, already decided, and did not pick you.

Where signals plug into the machine

A signal only creates value if something happens when it fires. That means signal work is not a data project, it is the input layer of a working system.

In practice, the signal feeds enrichment, enrichment feeds routing logic, and routing logic decides what asset a given account receives at what tier. We cover that architecture in GTM engineering, and the tiering it feeds in account-based marketing for B2B SaaS.

The most common organisational failure is a beautiful signal dashboard nobody acts on, because acting on it requires eleven manual steps. Signals that arrive faster than your team can respond to them are not an asset. They are a source of guilt.

Acting on a signal without being obvious

There is a real failure mode where signal-based outreach reads as surveillance. "I saw you visited our pricing page three times" is technically accurate and lands badly.

The rule that works: let the signal choose the timing and the topic, never the opening line.

If the signal is a job posting describing manual reconciliation work, do not mention the job posting. Open on the problem the posting implies, in their language, and let them make the connection. You demonstrate understanding without demonstrating monitoring.

Concretely:

• Bad: "I noticed you're hiring three ops analysts."
• Good: "Most teams at your stage hit a wall where reconciliation headcount grows faster than transaction volume."

Same insight, sourced from the same signal. One reads as informed. The other reads as watched.

What you build on top of that opening matters as much as the opening itself. The signal earned you the right to be relevant. What you send has to actually be relevant, which is a content problem covered in 1:1 ABM personalization.

So is intent data worth buying?

A defensible position:

Do not buy third-party intent data until you have instrumented first-party signals properly. Website de-anonymisation on your own high-intent pages, product usage events, and email engagement, all landing in one place where a human sees them. Most teams that "need better intent data" actually have excellent intent data that nobody has connected to anything.

Buy it when your first-party volume is genuinely too thin to fill a pipeline, and you are prepared to use it as a prioritisation input rather than a trigger. That is a legitimate use and it works.

Do not buy it because the dashboard was impressive in the demo. Every dashboard is impressive in the demo. The question to ask a vendor is not "how many accounts will light up" but "what share of accounts you flag as surging actually enter a buying process within 90 days". The answers, when vendors will give them, are more sobering than the demos suggest.

Frequently asked questions

What are B2B buying signals?

Observable events that raise the probability an account will enter a purchase process in the near term. They split into first-party signals, which happen on property you own, and third-party signals, which happen elsewhere. First-party signals are consistently more predictive.

What is the strongest buying signal in B2B?

Repeat pricing page visits from multiple people at the same account within a short window. It indicates a buying committee actively comparing, it is free, and no purchased data source matches its precision.

Is third-party intent data accurate?

Third-party surge data has structural limits regardless of vendor. The reader is anonymous, topic mapping is coarse, and detection lags the actual research. It works as a tiebreaker between otherwise equal accounts. It works poorly as a trigger to launch outreach.

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

First-party intent is behaviour on your own properties: pricing pages, docs, product usage, email engagement. Third-party intent is behaviour observed elsewhere, usually across a publisher network, and sold to you. First-party is free, precise and attributable. Third-party is broader, fuzzier and anonymous.

Are job postings a good intent signal?

Yes, when read properly. The signal is not that a company is hiring. It is that a specific role's responsibilities describe the problem your product solves, which means their current approach is failing publicly and with a date attached. It is one of the most under-used third-party signals in B2B.

How do you score intent signals?

Weight by signal strength, apply recency decay so signals lose value after roughly 14 days and expire near 60, multiply by the number of distinct people at the account, and double-weight when the person sits in the function that owns your problem. Treat two strong signals within 30 days as a top-priority trigger irrespective of score.

---

*NomiOS reads the signal layer for you. Paste a target company URL and it pulls firmographics, open roles and decision-makers, runs a live web research pass for the softer why-now signals, and returns a briefed account read plus a tailored account asset.*

[See how NomiOS works →](https://nomios.rzlt.io)