Predictive vs Manual
Predictive scoring (Enterprise) uses ML on your data. Manual scoring is rule-based. Start manual — you control the logic and understand the score. Add predictive later.
Score Properties
Multiple score properties for different purposes: fit score (firmographic), engagement score (behavior), composite score. Don’t reduce everything to one number.
Workflow Triggers
Scores drive workflow triggers: MQL conversion, sales notification, sequence enrollment. Set threshold bands, not hairline cutoffs.
Decay
Engagement score should decay. A contact engaged 18 months ago shouldn’t score the same as one engaged last week. Build decay into manual scoring.
Quarterly Review
Review score distribution quarterly. If 80% of leads score above MQL threshold, your threshold is too low. If 5%, too high. Calibrate against conversion data.