How Lead Scoring Works
Lead scoring operates on a two-axis model: behavioral score (what the prospect has done) and demographic or firmographic score (who the prospect is). Behavioral signals that add points typically include website visits to high-intent pages (pricing, demo, case studies), email clicks, content downloads, webinar attendance, and free trial activations. Demographic signals weight job title, company size, industry, and geography against the ideal customer profile. A VP of Engineering at a 200-person SaaS company visiting the pricing page three times in a week should score higher than a student at a university clicking the same page once. The combined score determines which leads receive immediate sales attention versus further nurture.
Why Lead Scoring Matters for B2B Marketing
Lead score decay is an often-overlooked component of effective scoring models. A contact who scored 85 points from a burst of activity six months ago but has had zero engagement since is not a hot lead — the score is stale. Decay rules automatically reduce behavioral score over time (typically -1 to -5 points per week of inactivity), ensuring the score reflects current engagement rather than historical peaks. Platforms like Marketo, HubSpot, and Pardot support configurable decay logic. Without decay, scores inflate over time and lose their predictive value for sales prioritization.
Lead Scoring: Best Practices & Strategic Application
Setting the MQL threshold — the score at which marketing hands a lead to sales — requires calibration against win rate data. The threshold is too low if sales reps are overwhelmed with leads that never convert; it is too high if qualified buyers are being held in nurture too long. The calibration process involves analyzing closed-won deals to identify the average score at first sales contact, then setting the threshold at a score where win rate exceeds a defined minimum (typically 15–25% for B2B). This data-driven threshold replaces the common practice of setting arbitrary round numbers (50, 75, 100) that have no empirical basis.
Agency Perspective: Lead Scoring in Practice
Predictive lead scoring extends rule-based models by using machine learning to identify patterns in closed-won data that humans would not detect manually. Platforms like 6sense, Salesforce Einstein, HubSpot's predictive scoring, and Infer analyze hundreds of firmographic, technographic, and behavioral signals to produce a conversion probability score rather than a manually configured point total. Predictive scoring typically outperforms rule-based models by 20–40% on MQL-to-SQL conversion rates because it captures non-obvious signal combinations. However, it requires a minimum of 200–500 closed deals in the training dataset to produce reliable predictions.
