64% of CMOs now carry direct responsibility for profits, not lead volume alone (IBM CMO Study 2025).
This shift changes what scoring must deliver: forecast accuracy, resource efficiency, and CFO confidence.
Your legacy scoring model lumps ICP fit with behavior into one number.
When a competitor's intern downloads ten whitepapers, they score higher than your ideal customer who visited once. Sales ignores the leads. Marketing defends the volume.
HubSpot sunset the legacy model on August 31, 2025, and back in the day, we had to use complex workflows (we wrote about it here) to achieve what is now a native feature. Here is also a great webinar by HubSpot User Groups on the topic.
If you haven't migrated, your scores are frozen and workflows depending on them no longer function.
The legacy HubSpot lead scoring model mixed two dimensions into one score.
Fit measures who they are—company size, industry, job title, revenue.
Engagement measures what they've done—form fills, email clicks, page visits.
Mixing these creates predictable problems.
A junior worker at a wrong-fit company downloads your entire resource library. Score rises.
Sales wastes a week chasing someone who can't buy.
A VP at your perfect-fit account visits your pricing page once. Score stays low.
Single-score models reward volume over signal quality.
65% of sales and marketing pros report team misalignment (Forrester 2024).
Much of that friction traces to lead scoring that can't explain why a lead qualified.
When you can't split fit from engagement, you can't forecast deals. CFOs notice.
HubSpot's new scoring splits concerns into three models.
[Diagram: Three-model setup showing Fit Score, Engagement Score, and Combined Score feeding into routing]
The matrix labels work as coordinates. Letters (A/B/C) = Fit tier. Numbers (1/2/3) = Engagement tier.
A1 means high fit, high engagement. C3 means low fit, low engagement.
Lead scoring for fit needs a step most teams skip: ICP definition.
Before setting fit scores:
Common fit criteria:
Award points for ICP matches. Deduct points for disqualifiers.
Example scoring weights:
| Criterion | Points |
|---|---|
| Revenue €10M-100M | +15 |
| Industry: SaaS or Services | +10 |
| Job Title: Director or VP | +10 |
| Country outside regions | -20 |
Set thresholds that create clear tiers:
Test against historical win rates. If B-tier leads convert like A-tier, adjust your thresholds.
Engagement scoring tracks actions that signal buying intent.
Weight events by conversion impact:
Six-month-old actions shouldn't count like last week's. Apply decay rates (10-20% monthly) to prevent "zombie leads" from appearing hot.
Caps prevent any single action from dominating the score.
[Screenshot: HubSpot engagement score setup showing event groups and decay settings]
Machine learning models rank sales-ready leads 77% more accurately than manual scoring (Martal Group 2025). Well-designed manual rules compete with AI when data is limited.
The 3x3 matrix creates nine routing categories:
| Category | Fit | Engagement | Action |
|---|---|---|---|
| A1 | High | High | Route to sales now |
| A2 | High | Medium | Sales follow-up |
| A3 | High | Low | Marketing nurture |
| B1 | Medium | High | SDR qualification |
| B2 | Medium | Medium | Nurture sequence |
| B3 | Medium | Low | Long-term nurture |
| C1 | Low | High | Partner channel or drop |
| C2 | Low | Medium | Drop |
| C3 | Low | Low | Drop |
Medium fit + high engagement creates ambiguity. The prospect shows interest but doesn't match ICP.
Route to SDR for qualification—not direct to AEs.
Low fit + high engagement often indicates:
High activity doesn't compensate for wrong fit.
Enterprise tier unlocks AI scoring that identifies patterns across hundreds of data points.
Requirements:
Organizations using AI scoring platforms see 138% greater ROI on lead efforts versus 78% without scoring (Martal Group 2025).
Start with manual scoring. Add AI after 6-12 months of conversion data.
HubSpot sunset legacy lead scoring on August 31, 2025. Historical scores remain as static data—they no longer update, and workflows depending on them stopped functioning. HubSpot may delete unused legacy properties in Q4 2025.
Migration timeline (if you haven't switched yet):
Common migration mistakes:
[Screenshot: HubSpot new lead scoring setup with fit and engagement models]
This isn't about meeting a deadline. It's about resolving years of scoring debt.
The migration forces the ICP definition work you've deferred.
HubSpot lead scoring works when you separate concerns and align your teams.
Your path forward:
The August 31, 2025 deadline has passed. If you're still on legacy scoring, migrate now—your data is already frozen.
Lead scoring tools don't solve scoring problems. Process design does.
The best model fails if sales doesn't trust it and marketing doesn't maintain it.