A buying signal is evidence, not a verdict.
That distinction matters because most signal programs fail in one of two ways. Teams either act on every alert and create noisy outreach, or they collect so much data that nobody knows what deserves action.
A useful scoring model sits between those extremes. It combines account fit with the strength, recency, frequency, and convergence of the observed events. The output is not “this company will buy.” The output is a practical next step: monitor, research, personalize, or follow up.
Start with fit before intent
No signal should rescue a poor-fit account.
Score ICP fit first:
- Company fit: market, size, geography, operating model, and relevant technology.
- Role fit: whether the person can own, influence, or experience the problem.
- Problem fit: whether the observed context connects to a problem your offer can solve.
- Commercial fit: whether the account is realistically serviceable.
A high-intent event from a company you cannot serve should remain disqualified. Signal-led selling adds timing to fit; it does not remove the fit requirement.
Score five dimensions
Use a simple 0–3 scale for each dimension. The exact numbers matter less than applying the same definitions consistently.
1. ICP fit
- 0: Outside the ICP.
- 1: Partial fit with major unknowns.
- 2: Good fit with one missing detail.
- 3: Strong company, role, and problem fit.
2. Signal strength
Strength describes how directly the event relates to a possible buying situation.
- 0: Broad activity with no clear business relevance.
- 1: Awareness-level context.
- 2: A meaningful trigger or repeated category engagement.
- 3: Direct engagement, implementation questions, or decision-stage behavior.
A funding announcement is usually a trigger, not proof of demand. A new executive hire can create a reason to investigate, but the initiative still needs validation.
3. Recency
- 0: Too old to shape a timely conversation.
- 1: Still informative, but no longer urgent.
- 2: Recent enough to guide prioritization.
- 3: Time-sensitive and useful now.
Different signals decay at different speeds. A role change may be useful for weeks. A response to a live conversation may require action the same day.
4. Frequency
- 0: One isolated, low-confidence event.
- 1: A single credible event.
- 2: Repeated related activity.
- 3: A sustained pattern over a short window.
Frequency should not reward duplicate data from several vendors. Count meaningful independent events, not repeated copies of the same source.
5. Convergence
Convergence asks whether multiple signals support one coherent hypothesis.
- 0: Events point in unrelated directions.
- 1: One plausible interpretation, but little support.
- 2: Several events align around one likely initiative.
- 3: Account, person, and engagement signals reinforce the same context.
Convergence is often more useful than any single “high-intent” label because it forces the team to explain why the evidence belongs together.
Turn the score into an action
A total score is useful only when it maps to a workflow.
0–5: monitor
The account may fit, but the context is weak, stale, or ambiguous. Keep it visible without forcing outreach.
6–9: research
There is enough evidence to investigate the account, likely initiative, and relevant buyer. The next step is context gathering, not sequence enrollment.
10–12: personalize
Fit and timing are credible. Draft a focused message around the underlying business situation. Avoid mentioning private-feeling tracking details.
13–15: prioritize or follow up
The evidence is strong and recent. Route it quickly, especially if there is already an open conversation or prior engagement.
These thresholds are a starting point. Calibration should come from real outcomes.
Add a confidence check
Before acting, ask three questions:
- Source: Do we understand where the event came from?
- Attribution: Are we confident it belongs to the right company or person?
- Interpretation: Is there another reasonable explanation?
If source quality or attribution is weak, lower the action level even when the numerical score is high.
Measure by signal and play
Do not evaluate the program only on total messages or total replies.
Track:
- accepted and rejected matches
- false-positive rate
- time from signal to action
- replies and positive replies
- meetings and opportunities
- pipeline by signal type
- performance of signal combinations
This creates a feedback loop. Signals that produce activity without qualified outcomes should lose weight or leave the model entirely.
Keep the model explainable
A score should help a seller understand the decision. It should not hide judgment behind a number.
For every prioritized account, the team should be able to explain:
- why the account fits
- what changed
- why the context matters now
- what is still uncertain
- which next action is appropriate
NetworkHQ’s broader buying-signals library covers signal types, likely strength, missing context, and recommended actions. Use it as a reference, then tune the weights around your own market and conversion data.