How to score inbound leads with AI
Classical lead scoring is rule-based and inflexible. AI-augmented scoring handles free-text fields (challenge descriptions, custom requirements) and free-form intent signals. 4-component model below.
Tools you'll need
Steps
- 1
Define the 4 scoring components
FIT (40 pts max — firmographic fit). INTENT (30 — explicit buying signals). ENGAGEMENT (20 — behavioural). URGENCY (10 — time-based). Total 100.
- 2
Allocate FIT points
Company size 10-500 employees (+15). Revenue $1M-$50M (+10). Industry in your top-7 ICP (+10). Title is decision-maker (+5).
- 3
Allocate INTENT points
Filled audit form vs. newsletter (+15). Reported manual hours ≥30/week (+10). Wrote >100 chars in challenge field (+5). Use AI to score the challenge field for specificity vs. generic.
- 4
Allocate ENGAGEMENT + URGENCY points
Visited ≥3 pages same session (+5). Viewed pricing (+5). Return visit within 7 days (+5). Mentioned urgency keywords (+5). Currently on a competitor's tool (+5).
- 5
Set routing thresholds
85+ HOT: 5-min SLA, Slack-ping AE. 65-84 WARM: 4-hour SLA, SDR queue. 40-64 NURTURE: auto-sequence. 0-39 FUTURE: archive + quarterly re-score.
- 6
Calibrate against historical closed-won data
Pull last 90 days of closed-won deals. Calculate their score at the time they entered. If your model says >85 for 80% of them, you're calibrated. If not, tune weights.
- 7
Re-score every quarter
Closed-won data accumulates → re-tune weights monthly first 3 months, then quarterly. Track which components predict best — typically FIT does, INTENT second.