B2B· SaaS Marketing

B2B Intent Signals: Turn Buyer Activity Into the Right Next Action

Founder, Grow Predictably

16 min read3,108 words
B2B Intent Signals: Turn Buyer Activity Into the Right Next Action

TL;DR: A B2B intent signal is evidence, not a verdict. Test every signal on four checks (fit, specificity, recency, convergence), let the weakest consequential check cap the response, and record each usable signal on a 14-field card that ties evidence to an action and a qualified outcome.

Key takeaways

  • Activity is not intent. One click can be research, curiosity, a student, a competitor, or a real buying decision.
  • Name the buyer decision first, then pick the evidence. Don’t start from the data vendor.
  • Four checks, no score. Fit, specificity, recency, and convergence produce a written judgment, not a number.
  • Match the response to the evidence: observe, nurture, investigate, route, or invite.
  • Judge signals against a qualified outcome defined in your CRM. Outcomes after a response don’t prove the response caused them.

Sales teams are told to “act on intent.” The harder question is what to do when a signal fires. Act too hard on weak evidence, and you damage trust. Wait for certainty, and you miss the window.

This guide gives you a way to take one real signal, test it, and choose a proportional response. It’s built on a method, not a tool.

B2B intent signals are evidence, not a verdict

A B2B intent signal is an observed behavior that changes the probability that an account is working on a buying decision. It does not prove the account is ready to buy, and it doesn’t show which person wants to talk to sales.

So when a signal fires, ask how much confidence the evidence has earned. Then take the smallest action that fits.

Four terms keep the reasoning clean:

  • Activity: what you observed, such as a page view, webinar registration, or product event.
  • Interest: your inference about what the person wanted to learn.
  • Fit: whether the account and person resemble customers you can serve well.
  • Readiness: whether a purchase decision is active and the buying group is prepared to hear from a vendor.

Activity is observed. Account fit is checkable against your records. A person’s role in the buying group is partly inferred. Interest and readiness are inferences, and any of them can be wrong.

Anonymous research widens that gap. In 6sense’s 2026 study, 35.2% of surveyed buyers used their real contact information on a form and buyer teams initiated 82% of first vendor contacts. Both figures describe that study’s respondents, so they show a pattern worth planning around, not a rate to expect in your own market.

A team that waits for a form will miss earlier research. A team that reacts to every anonymous click will chase noise.

Start with the buyer decision, not the data vendor

Programs often start with a tool. Someone buys a data feed, then sales is asked to find a use for it. Reverse the order. Name the decision the account may be making, because the decision determines which evidence is relevant.

A decision is specific:

  • Decision: “The account is evaluating a replacement for its current reporting tool.”
  • Topic: “The account is interested in reporting.”

A topic can match accounts for reasons that have nothing to do with buying. A decision gives you something to test. Which behaviors, and which roles, would show it? If you can’t observe that evidence, more data of the same kind won’t reduce your uncertainty.

I work from one habit: diagnose before you treat. Name the stage that is capping growth, then apply tools there. Map the signal to the buyer’s decision stage first, so you know whether the activity is early learning, active evaluation, or a need after purchase.

I supported an account-based marketing initiative by leading parts of a 6sense taxonomy project. My operating judgment is that a taxonomy can label activity neatly and still fail as a decision system. Labels become useful only when the team knows which buyer decision they inform and what action they permit.

Before you collect another signal, check for four blanks:

  1. The buyer decision
  2. The owner
  3. The strongest action the evidence permits
  4. The qualified outcome that will test the logic

A blank in any of these is the constraint. More scores, alerts, or dashboards only make the gap harder to see.

Require two things up front: a named owner and an allowed action. A signal with no owner is a chart. A signal with no allowed action starts an argument every time it fires. Settle both before the signal exists, so the argument happens once, calmly, not in the middle of a deal.

Use a four-part test before you act

Ask four questions of every signal: fit, specificity, recency, convergence. The test produces a short written judgment of what you know, what you’re assuming, and the strongest response that judgment supports.

This guide gives no weights or thresholds. The right ones depend on your sales cycle, your market, and the outcomes your CRM can verify.

Fit

Is this the right account, and is the observed person relevant to the buying group?

Check the account against your ideal customer profile, customer status, and any open opportunity. Check the person’s role and function. A student, a competitor, an existing customer’s support user, and a recruiter can all generate activity that looks like evaluation.

Fit comes first because it’s the cheapest question to answer. A poor answer here can end the analysis.

Specificity

How close is the activity to the actual decision?

A broad educational article says little about a purchase. An integration guide, security page, or pricing page says more, because those pages answer questions people ask while evaluating. Even so, one visit doesn’t prove readiness, since a pricing page also serves the merely curious.

Judge specificity by distance from the decision you named, not by channel or content type.

A contact-sales request gets more specific when the form captures information that narrows the decision, such as role, need, or timing. Every question should earn its place by changing qualification, routing, or follow-up. Otherwise, it adds friction without adding insight. I applied this reasoning in 2024 while working with a marketing automation team on lead form optimizations for contact-sales forms.

Recency

Evidence ages.

Write a decay window for each signal type on your card. Set it from your own sales cycle, and let the signal fall back to observation when the window passes without new activity. Don’t borrow another company’s number, because their cycle isn’t yours.

To choose your own window:

  1. Compare CRM accounts that reached the qualified stage with similar accounts that didn’t.
  2. For each, mark the first relevant activity and either the qualification date or the end of the observation period.
  3. Compare how often activity is followed by qualification across different windows.
  4. Set a provisional window, then revise as more cases arrive.

Convergence

Do multiple independent signals point the same way?

Independence matters. Five page views by one person in one session are one signal repeated, not five. Convergence looks like a fitting account, a second person from the same buying group, and behavior close to the decision, with CRM context that doesn’t contradict any of them.

Contradiction counts as evidence too. A closed-lost opportunity from last month, an unsubscribe, or a note from a salesperson who already spoke with the account belongs in the judgment.

When the four tests disagree, the weakest consequential answer caps the response. A fitting account with specific but stale activity doesn’t earn stronger action until fresh evidence arrives. A recent, specific action from a poor-fit account still fails the fit test.

Read the account, not one click

A contact’s activity is one person’s behavior. A B2B purchase involves a buying group, and one member can research alone, advocate without authority, or read about a problem the company has already decided not to solve. Why one person’s interest is not group agreement is its own subject. The operating rule is short:

  • Treat contact activity as evidence about that contact.
  • Treat buying-group movement as a separate claim that needs its own evidence, such as a second role engaging or an existing opportunity advancing.

First-party vs. third-party signals

  • First-party signals are behaviors you observed on properties you control: your website, product, campaigns, and CRM. You can inspect them, tie them to a record, and check them against what happened next.
  • Third-party signals describe activity somewhere else, reported by a provider. You didn’t observe the behavior, so ask what you can verify about how the provider reached its conclusion.

Use a third-party signal as a reason to investigate. Use first-party evidence and CRM context to decide whether a stronger step is justified.

CRM context keeps either type honest. Before anyone acts, check whether the account is a customer, has an open opportunity, was closed-lost recently, or has a contact who asked not to be contacted. A strong signal from an account already in a sales conversation is not a lead. It’s a reason to tell the owner of that conversation.

Identity has limits. Anonymous activity doesn’t identify an individual, and a tool that ties an anonymous visit to a company is making an inference that can be wrong. Don’t attach anonymous activity to a named person, and don’t write outreach that implies you watched what they did.

Privacy limits are part of the evidence. The action a signal allows shrinks with them. This guide doesn’t cover privacy law, so settle those questions with your own counsel before you automate.

Match the response to the strength of the signal

Pair the four-part test with a response ladder that keeps weak evidence from triggering strong outreach. Each rung asks for more fit, specificity, recency, and convergence than the rung above it in the table. The ladder is a working convention, not a maturity model, and your sales cycle decides how many rungs you need.

#ResponseEvidence it needsAllowed actionStop or suppress when
1ObserveA signal with unresolved fit, low specificity, or an anonymous sourceRecord the signal and watch for a second independent oneThe decay window ends with no new activity
2NurtureGood fit with broad or early behaviorOffer relevant educational content through a channel the person already choseThe person unsubscribes or stops engaging
3InvestigateGood fit plus specific behavior, or two signals that do not yet agreeResearch the account, check CRM history, and look for the buying groupThe research shows a customer, a competitor, a poor fit, or an unexplained contradiction
4RouteGood fit, specific behavior, and recent activity from more than one independent signal that agreeAssign the account to a named owner with the evidence attachedThe owner finds no real opportunity, and the case is logged as a false positive
5Invite a sales responseRoute-level evidence plus a reason the person would welcome contactOffer a relevant conversation or tailored resource to someone in the buying groupThe person declines, ignores the invitation, or has asked not to be contacted

The cost of being wrong rises as you move down the table. Observing a weak signal is cheap. An intrusive sales message is not. Gartner reported that 73% of surveyed B2B buyers actively avoid suppliers that send irrelevant outreach, in a survey of 632 B2B buyers conducted from August to September 2024. A buyer who avoids you takes a possible future deal out of play.

The ladder also guards against the opposite error. Certainty doesn’t arrive, so the lower rungs let you act early at low cost with relevant content and quiet research. The upper rungs stay reserved for evidence that has earned them.

Three rules keep the ladder from drifting:

  1. One owner per rung. Two teams acting on one signal means the buyer gets two messages.
  2. Suppression rules for customers, open opportunities, recently closed-lost accounts, and anyone who opted out.
  3. A stop condition on every action, as in the last column, so a response that isn’t working ends instead of repeating.

Connect intent signals to qualified pipeline

Define the qualified outcome in your CRM before you score any activity. Then keep four things separate:

  • Engagement
  • A qualified lead
  • An opportunity
  • Closed-won revenue

Engagement is evidence. The other three are outcomes that test whether the evidence meant anything. Calling engagement pipeline hides the difference.

Feedback arrives late. A signal can fire in one quarter while its opportunity closes in a later one. Compare signals and outcomes over a period that matches your sales cycle, and don’t recalibrate from a few weeks of results. A CRM outcome after a response shows two things happened in sequence, not that the response caused the outcome.

Two practical rules:

  • Sparse outcomes: pool a longer period or leave the rule provisional.
  • Different stage definitions across teams: agree on one qualified stage and document it first.

LinkedIn offers one inspectable example. Its help documentation says LinkedIn can receive offline qualified-lead events through CRM Sync or Conversions API and use them for qualified-lead optimization. The principle outlives the platform: a qualified-lead definition you own can serve as feedback in place of optimizing for lead volume alone.

Its Companies Hub shows the same separation, since engagement views are distinct from CRM-based revenue views that include pipeline and closed-won revenue. The paid-media side of that loop is covered in AI-powered B2B ad campaign management.

What I’ve seen on the outcome side

I worked on the outcome side of this loop inside a B2B software company. In separate cross-functional work, I collaborated with automation and legal teams on projects that used data analysis and AI-assisted research to find which website behaviors could serve as leading indicators for the employer’s qualified-lead measure, called AQL internally.

I kept asking how to align the data for leading and lagging indicators. For this method, substitute your own defined stage, such as MQL.

  • AQL, first half of 2025: volume across Contact Sales, Chat, and Product Showcase streams grew 58% year over year. The Contact Sales stream grew 222% year over year on the same measure.
  • Q4 2024: I worked on a multi-team project that planned and executed an optimized web experience for prospective customers contacting sales. Pipeline from the Contact Sales form increased 68% in that period.

These figures give period context for those measures. They do not isolate the effect of any signal method or form change, and nothing here claims that one produced them.

The lesson from both projects: a taxonomy isn’t finished when every behavior has a label. It’s finished when a team can connect each behavior to a buyer decision, a named owner, a bounded response, and a qualified outcome. The card below carries all four.

Build a signal-to-action card

Complete one card per signal you plan to act on. If you can’t complete a field, you’ve found the gap in your logic before it became an outreach mistake.

The card has three groups: Evidence (Buyer decision through Decay window), Action (Owner, Allowed action, Exclusions, Stop rule), and Feedback (Qualified outcome, Review date). The worked example is generic and invented only to show the format. It is not a client result.

[Approved image 2 goes here: the blank 14-field card.]

GroupFieldWhat to writeWorked example
EvidenceBuyer decisionThe decision the account may be makingEvaluating a replacement for its current reporting tool
EvidenceSignalThe observed behaviorA second person from the account views the integration guide after a colleague viewed pricing
EvidenceSourceWhere it was observed and how it was matchedFirst-party web analytics matched to known CRM contacts, not anonymous traffic
EvidenceFit ruleThe account and role criteriaIn the target segment, not a customer, no open opportunity, relevant function
EvidenceSpecificityDistance from the decisionIntegration and pricing pages sit closer to evaluation than educational articles
EvidenceRecencyWhen the behavior happenedBoth visits fall inside the decay window
EvidenceConvergenceWhich independent signals agreeTwo roles, two late-stage pages, and no contradicting CRM note
EvidenceDecay windowHow long the signal stays actionableA number of days set from your own sales cycle
ActionOwnerThe person who actsA named sales development representative for the segment
ActionAllowed actionThe most the owner may doRoute to the named representative with the evidence attached. The representative may verify account context and look for a reason the person would welcome contact, but may not contact the person on route-level evidence alone
ActionExclusionsWho is never contactedCustomers, open opportunities, recently closed-lost accounts, anyone who opted out
ActionStop ruleWhat ends the actionThe owner finds no opportunity, the evidence falls outside the decay window, or an exclusion applies
FeedbackQualified outcomeThe CRM result that countsThe account reaches the qualified-lead or opportunity stage as your CRM defines it
FeedbackReview dateWhen the signal is judgedThe date the owner sets when routing the account

Use the review date to look both ways:

  • False positives: accounts you routed that didn’t qualify. Ask which of the four tests failed.
  • Missed opportunities: accounts that qualified without any earlier signal. Ask what evidence you weren’t collecting.

Change the card based on both lists, and record the change so the next review compares like with like. Because a CRM outcome doesn’t prove causation, treat each review as a reason to adjust the logic, not as a verdict on a tool.

Know when the signal is too weak to use

Stop, or stay at observation, when the evidence is:

  • Anonymous: it can’t support outreach to a person.
  • Stale: it has decayed past its window.
  • Sparse: one visit doesn’t converge with anything.
  • Contradictory: strong behavior from an account that just closed-lost needs an explanation before it needs a response.
  • Limited by privacy: privacy limits set a ceiling on the action, whatever the signal says.

Don’t let a vendor score replace judgment. A score is another signal and faces the same four questions. Can you see what it was built from? Does it fit your accounts? Is it recent? Does it agree with first-party evidence? If you can’t answer, it earns observation at most.

If evidence is sparse because few visitors convert at all, the constraint may sit in the conversion path, not the signal. Diagnose the conversion path before you add acquisition activity to produce more signals.

A generic misfire: an account downloads one educational guide, and an automation sends a sales sequence the next day. Fit was never checked, specificity was low, and nothing converged. The ladder supports observation or a relevant resource at most. Recovery means narrowing the action, adding CRM status to the check, and defining the qualified outcome you’ll judge the change against.

Every signal ends in one of four decisions:

  • Observe when the evidence is incomplete but could grow.
  • Test when a small, low-risk action could produce better evidence. That covers nurture (a relevant resource) and investigate (quiet account research).
  • Route when fit, specificity, recency, and convergence agree. That includes inviting a sales response when the contact would welcome it.
  • Stop when the evidence contradicts the action or a limit rules it out.

Acting on weak evidence damages trust, and waiting for certainty can miss a real window. The ladder and the card exist to manage that tension.

Find My Growth Gap is a starting point for locating the stage that is capping growth in your own funnel.

About the author

Brian K Shelton, Founder of Grow Predictably
Brian K SheltonFounder & Growth Strategist, Grow Predictably

DigitalMarketer Certified Partner (since 2020). 15+ years across in-house and agency marketing roles.

Ready to see your real growth gap?

Two minutes in the Growth Gap Scan and you walk away knowing the one thing capping your pipeline. Free, no obligation.

Find My Growth Gap