wikis / Lead Generation / wiki / concepts / lead-scoring.md view as markdown report a mistake

Lead Scoring

Lead scoring assigns points or categories to selected signals so teams can prioritize records consistently. It is a ranking mechanism, not proof that a person wants to buy.

Signal groups

Separate two questions:

• Fit: Does the account or person resemble the target customer?
• Engagement: Has the person taken actions that plausibly indicate interest?

Fit signals may include industry, company size, geography, role, or technical requirements. Engagement signals may include form submissions, product actions, event attendance, repeated high-intent page visits, or replies. A combined view can prevent highly active poor-fit records and inactive good-fit accounts from being treated identically.

Model design

Assign more weight to signals closer to meaningful intent. Cap repeated weak actions so they do not overwhelm stronger evidence. Use negative points for known incompatibility or declining engagement, and consider time decay when old behavior should matter less.

Define thresholds from observed outcomes, then validate them against qualified conversations, opportunities, and customers. Review false positives and false negatives with sales. Different products, regions, or motions may require separate models.

Keep the model interpretable. A user should be able to explain why a record received a score and what action the threshold triggers. Do not encode protected or sensitive attributes, unreliable proxies, or hidden inferences merely because they correlate with conversion.

Scoring can trigger routing, nurture, or review, but lead qualification remains the broader judgment. Document changes so historical performance and campaign tracking and attribution are not misread after a model revision.