Lead Scoring Framework
Build a data-driven lead scoring model with this framework. Covers fit scoring, intent signals, and engagement tracking for B2B sales teams.
Lead Scoring Framework
Not every lead deserves the same attention. This framework helps you build a scoring model that separates high-intent buyers from tire-kickers, so your sales team focuses on the prospects most likely to convert.
Scoring Dimensions
Your lead score should combine three dimensions, each weighted by importance to your business.
1. Fit Score (0-40 points) How well the prospect matches your ideal customer profile.
| Criteria | Points |
|---|---|
| Company size matches ICP | 0-10 |
| Industry match | 0-10 |
| Job title/seniority match | 0-10 |
| Geography match | 0-5 |
| Tech stack alignment | 0-5 |
2. Intent Score (0-35 points) Signals that the prospect is actively looking for a solution.
| Signal | Points |
|---|---|
| Visited pricing page | +10 |
| Downloaded content asset | +5 |
| Attended webinar | +5 |
| Job posting matching your category | +8 |
| Recent funding round | +7 |
3. Engagement Score (0-25 points) How actively the prospect interacts with your outreach.
| Action | Points |
|---|---|
| Opened 3+ emails | +5 |
| Clicked email link | +8 |
| Replied to outreach | +10 |
| Visited website from email | +5 |
| Connected on LinkedIn | +5 |
Lead Tiers
| Tier | Score Range | Action |
|---|---|---|
| Hot | 70-100 | Immediate sales outreach |
| Warm | 40-69 | Nurture sequence + SDR follow-up |
| Cool | 20-39 | Marketing nurture only |
| Cold | 0-19 | No action — re-evaluate quarterly |
Implementation Steps
- Define your ICP criteria and weight each factor
- Map intent signals available from your data sources
- Set up engagement tracking across email and website
- Score your existing database to calibrate thresholds
- Review and adjust scoring weights monthly based on conversion data
Scrapine’s AI lead scoring automatically enriches contacts with firmographic and technographic data, giving you the raw inputs needed to power your scoring model without manual research.