The most predictive buyer intent signals are multi-signal corroboration, late-stage first-party behavior (pricing, demo, and repeat product research), competitor-comparison research, and relevant decision-maker changes at ICP-fit accounts. These correlate hardest with opportunity creation; for closed-won specifically, budget and funding events plus multi-threaded engagement matter most. Broad, single-source topic surges are the weakest predictors on their own.

The Most Predictive Signals: The Short Answer

  • For opportunities: corroborated behavior + event, late-stage first-party visits, and competitor research predict best.
  • For closed-won: funding/budget events, multi-threaded engagement, and expansion usage carry the most revenue signal.
  • The floor: predictiveness is lift over an account's own baseline, not raw volume.
  • The trap: a broad, single-source topic surge looks busy and predicts almost nothing on its own.

How Signal Predictiveness Is Measured

Predictiveness is not "does this signal feel important." It is a measurable relationship between a signal firing and a downstream revenue outcome. Two numbers matter, and they are not the same:

  • Correlation with opportunity creation. Of the accounts that fired this signal, what share became a qualified opportunity within a defined window (say 30–90 days), compared with a matched control that did not fire it? That ratio is the signal's lift for pipeline.
  • Correlation with closed-won. Of those opportunities, what share actually closed, and at what cycle length and deal size? A signal can reliably create pipeline that never closes — high opportunity lift, low closed-won lift. That gap is where most teams overspend.

Three rules keep the measurement honest:

  1. Measure lift over the account's own baseline, not raw activity. A big enterprise researches everything; a lift number normalizes for that. This is the same normalization behind how intent data is collected and scored.
  2. Hold a control cohort. Without accounts that did not get the signal, you cannot separate the signal's effect from sales effort or seasonality.
  3. Attribute to the outcome, not the touch. Credit the signal that preceded the opportunity, then re-check whether it also preceded the win — the two answers rank signals differently.

The Most Predictive Buyer Intent Signals, Ranked

Here are the signals that survive that measurement, ordered by how much they lift the odds of a real outcome. The two outcome columns are rated separately on purpose — the order changes depending on whether you care about creating pipeline or closing it.

Rank Signal Predicts opportunities Predicts closed-won Why it holds up
1 Multi-signal corroboration (behavior + discrete event + ICP fit) Very high Very high Stacking independent signals cancels out each one's noise; the combined precision beats any single input.
2 Late-stage first-party behavior (pricing, demo request, repeat product/docs reads) Very high High Directly observed evaluation intent; perishable, so it predicts near-term pipeline strongly.
3 Competitor / alternative comparison research High High Comparing vendors is a late-funnel act — the account has a budget line and a shortlist.
4 Relevant decision-maker change into a buying role High Moderate–high A new owner reorganizes tooling and opens opportunities fast; closing still depends on their mandate.
5 Budget / funding events Moderate High Weaker at starting deals but strong at finishing them — capital removes the top reason deals stall.
6 Narrow keyword-level research spike above baseline Moderate–high Moderate Specific enough to map to a job-to-be-done; still an inference, not an observed action.
7 Expansion / usage signals in the install base Moderate High (expansion) For customers, product usage is the single best predictor of a closed expansion.
8 Broad category topic surge (single source) Low–moderate Low Too coarse and too noisy alone; useful only as corroboration for a stronger signal.

The single most important row is the top one. No individual signal in rows 2–8 is as predictive as two or three of them stacking at the same account inside the same window. For the full catalog of what each raw signal is and where it comes from, see the field guide to B2B intent signals.

Which Signals Best Predict Closed-Won (Not Just Opportunities)

Opportunity creation and closed-won reward different signals, and conflating them is the most expensive mistake in intent scoring.

  • Signals that create opportunities are about attention and access: a fresh decision-maker, a demo-page visit, a keyword spike. They open doors. But an opened door with no budget behind it produces pipeline that ages out.
  • Signals that predict closed-won are about capacity and consensus: a funding round or budget event (the money exists), multi-threaded engagement across several stakeholders (the buying committee is real, not one champion), and — for existing customers — expansion usage that proves value is already landing.

The practical read: a relevant hire is a great opportunity signal but a mediocre closed-won signal, because one enthusiastic new leader can still be overruled by a committee that never engaged. Funding is the reverse — it rarely starts the conversation, but it strongly predicts that a started conversation finishes. When you can see engagement spread from one contact to several at an account, treat that spread itself as a top-tier win predictor; single-threaded deals under-close no matter how hot the original signal was. This is also why early research versus in-market intent belong in different buckets: early research predicts eventual pipeline, in-market behavior predicts this quarter's wins.

Why Low-Predictive "Vanity" Signals Mislead

Weak signals do not just under-predict — they actively mislead, because they are abundant and feel like progress.

  • Broad category surges fire for entire industries at once. They correlate with the market, not with a specific account's readiness, so their lift over baseline is close to zero.
  • Generic email opens and page views measure your marketing's reach, not the buyer's intent. Opens inflate with send volume and tell you little about who will take a meeting.
  • Single-source, un-corroborated third-party topic data carries the most false positives. Used alone it manufactures a busy queue of accounts that look active and convert poorly.
  • High raw volume at big accounts looks like intent but is often just size. Without baseline normalization, your "hottest" list is a list of your largest prospects, not your readiest ones.

The tell is always the same: a vanity signal ranks accounts you already knew about and rarely surfaces a surprising, winnable one. If a signal cannot beat "just call the biggest logos," it is not predictive — it is decoration.

How to Operationalize the Predictive Signals

Knowing which signals predict is only half the job; the workflow has to route on that ranking.

  1. Gate on ICP fit first. A perfectly predictive signal at an account that can never buy is still noise. Filter, then score.
  2. Score on corroboration, not count. Reward accounts where a behavioral signal, a discrete event, and fit line up — not accounts with the most raw hits. Multiply the factors; do not add them.
  3. Split your scoring by outcome. Keep an opportunity score and a closed-won score. Send the opportunity-heavy accounts to reps for fast outreach; flag the closed-won-heavy ones (funding, multi-thread) for your best closers and tighter forecasting.
  4. Attach a verified contact to every predictive signal. A signal you cannot email is a story, not an alert. Use how to read Trigger Signals to turn a ranked, corroborated signal into a contactable person at the account.
  5. Re-measure quarterly. Predictiveness drifts as your market and product change. Recompute each signal's lift against fresh closed-won data and demote anything that stopped correlating.

For turning that ranking into a daily, capped, workable queue, see how to prioritize buying signals for outbound.

Frequently Asked Questions

Which buyer intent signals are most predictive of new opportunities?

The most predictive signals for opportunity creation are multi-signal corroboration (a behavioral signal plus a discrete event at an ICP-fit account), late-stage first-party behavior such as pricing- and demo-page visits, competitor-comparison research, and a relevant decision-maker moving into a buying role. Each shows near-term evaluation intent or verifiable change, which lifts the odds of a qualified opportunity well above an account's baseline.

Which intent signals best predict closed-won deals specifically?

Closed-won is best predicted by capacity-and-consensus signals rather than attention signals: budget or funding events (the money exists), multi-threaded engagement across several stakeholders (a real buying committee), and — for existing customers — expansion usage in the install base. A signal can create pipeline that never closes, so score these win-predictive signals separately from opportunity-predictive ones.

How is the predictiveness of an intent signal measured?

Measure it as lift over baseline against two outcomes. First, the share of accounts firing the signal that become opportunities within a defined window versus a matched control that did not fire it. Second, the share of those opportunities that reach closed-won, with cycle length and deal size. Normalize for the account's own baseline activity and hold a control cohort so you isolate the signal from sales effort and seasonality.

Why are broad topic-surge signals weak predictors?

Broad category topic surges fire for whole industries at once, so they correlate with the market rather than a specific account's readiness, and their lift over baseline is near zero. Used alone they generate false positives and a busy queue that converts poorly. They only earn their place as corroboration for a stronger, more specific signal.

Can a single intent signal reliably predict a deal?

No. Any single signal is a probability, not proof, and the most reliable predictor is corroboration — a behavioral or research signal, a discrete event, ICP fit, and a verified contact lining up at the same account inside the same window. Acting on one raw signal as a stand-alone call list is how teams build expensive, low-converting pipeline.

Sources

Next Steps

Predictive signals pay off when the ranking meets a verified contact and a corroborating event, so a rep can act on the readiest account instead of the biggest one. Start with the full menu of clues in the field guide to B2B intent signals, learn to weight them into a daily queue with how to prioritize buying signals for outbound, then separate the early-pipeline signals from the in-quarter ones with early research versus in-market intent.