Why Most Lead Scoring Fails
Most B2B lead scoring models fail for one reason: they over-weight behavioral signals. A competitor researching you will look like your best lead. An intern downloading whitepapers will outscore a VP who viewed pricing once.
The fix is a two-axis model: demographic fit (who they are) combined with behavioral engagement (what they do). Neither alone is sufficient.
Axis 1: Demographic Scoring (The Grade)
Demographic scoring answers: "Is this someone we can actually sell to?"
Assign a letter grade (A-F) based on ICP fit:
| Grade | Definition | Example |
|---|---|---|
| A | Perfect ICP match | VP Marketing at 500-5000 employee B2B SaaS |
| B | Good fit, minor gaps | Director Marketing at 200-500 employee tech |
| C | Possible fit, needs validation | Marketing Manager at enterprise (may not be decision-maker) |
| D | Poor fit, low priority | Right title but wrong industry or too small |
| F | Never sell to | Students, competitors, personal emails, agencies |
Demographic Criteria to Score
Company Fit
Industry, employee count, revenue, geography, tech stack
Contact Fit
Job title, seniority, department, decision-making authority
Axis 2: Behavioral Scoring (The Score)
Behavioral scoring answers: "How engaged are they right now?"
Assign points for actions that indicate buying intent:
| Action | Points | Rationale |
|---|---|---|
| Requested demo/contact | +30 | Explicit intent signal |
| Viewed pricing page | +20 | Evaluating cost = late stage |
| Visited 3+ product pages | +15 | Active research |
| Downloaded case study | +10 | Evaluating proof points |
| Attended webinar | +10 | Time investment |
| Opened 3+ emails | +5 | Consistent engagement |
| Downloaded TOFU content | +5 | Early awareness (low weight) |
| Unsubscribed | -20 | Negative signal |
| No activity 30+ days | -10 | Decay for inactivity |
Points should decay over time. A pricing page visit 6 months ago isn't the same as yesterday. Implement weekly or monthly decay (e.g., -5 points per 30 days of inactivity, cap at -30).
Combining Grade + Score into MQL
The magic is in the combination. Here's a qualification matrix:
| Grade | Score 0-30 | Score 31-60 | Score 61+ |
|---|---|---|---|
| A | Monitor | MQL | Hot MQL |
| B | Monitor | Monitor | MQL |
| C | Nurture | Monitor | Monitor |
| D/F | Disqualify | Disqualify | Disqualify |
Key insight: An A-grade lead with moderate activity (31+) should MQL. They're a perfect fit showing interest. Don't make them jump through hoops.
Conversely: A C-grade with 100 points shouldn't auto-MQL. High activity from a poor-fit contact is often a competitor, student, or non-buyer.
Implementation Tips
1. Start Simple, Then Iterate
Don't build a 50-criteria model on day one. Start with 5 demographic attributes and 5 behavioral triggers. Add complexity as you learn what predicts conversion.
2. Validate Against Closed-Won
Pull your last 50 closed-won deals. What did those contacts look like before they converted? Build your model around that, not theoretical ICPs.
3. Review Monthly
Lead scoring isn't "set and forget." Review MQL→SQL conversion rates monthly. If A-grade leads are converting at the same rate as C-grades, your grading is broken.
4. Get Sales Buy-In
The best scoring model is useless if sales doesn't trust it. Involve them in defining grades and thresholds. Their feedback on lead quality is your calibration data.