Why most lead scores are noise
The typical lead score adds a point for an email open, a few for a webinar, ten for a demo request, and calls it a day. The result correlates with engagement, not intent to buy. Reps chase warm-feeling leads that never had budget while real buyers sit unscored in the queue.
Score on two axes, not one
A single number hides the most important distinction. Split your score into:
- Fit — does this account look like a customer who succeeds? Industry, company size, tech stack, region.
- Intent — are they actually in a buying motion right now? Pricing-page visits, repeated logins, multiple stakeholders engaging.
Plot leads on a 2x2. High fit + high intent goes to sales today. High fit + low intent goes to nurture. Low fit gets disqualified no matter how engaged they look.
Weight behaviors by what predicts revenue
- Pricing page, twice in a week — the single strongest signal in most B2B funnels.
- Multiple contacts from one domain — a buying committee is forming.
- Demo request — high, but only when fit is also high.
- Generic content download — near zero. Stop scoring whitepaper grabs.
Decay matters
Intent is perishable. A pricing-page visit three months ago means nothing. Apply a decay curve so scores fall off if the prospect goes quiet — otherwise your "hot" list fills with stale accounts.
Validate against closed-won
Pull your last 100 closed-won deals and check: did your model actually score them highly before they bought? If not, your weights are guesses. Re-fit them on real outcomes every quarter.