Lead Scoring Model Template (Free B2B Framework That Sales Will Actually Use)

Lead scoring in B2B has a credibility problem. Most models are point systems that started reasonable, got tuned by committee, and ended up generating scores nobody trusts. Sales ignores the scores. Marketing keeps adjusting the weights. The handoff process keeps breaking.

The fix is a simpler model. Four dimensions. Point values that map to real buyer behavior. A scoring threshold that gets reviewed once a quarter, not constantly tweaked.

Here is the template I use with B2B teams. Free to copy and adapt.

The four dimensions

Every B2B lead has four orthogonal scoring dimensions. Score them separately. Combine them at the end. The separation is what makes the model debuggable.

Dimension 1: Fit

Does the lead match your ICP? Pulled directly from the ICP and persona worksheet. Company size, industry, role, geography.

Score 0 to 100. 100 means a perfect match. 0 means they should never have entered the funnel. Most leads land between 40 and 80.

Dimension 2: Intent

How clearly is the lead signaling they are looking? Pricing page views. Demo request. Comparison page reads. Multiple sessions in a short window.

Score 0 to 100. 100 means they have requested a demo or asked for pricing. 0 means they downloaded one piece of content six months ago and have not been back.

Dimension 3: Engagement

How deep have they engaged with your content and team? Newsletter opens. Webinar attendance. Multiple email clicks. LinkedIn follows from your founder.

Score 0 to 100. 100 means active engagement across multiple channels over the past 90 days. 0 means a single interaction that never repeated.

Dimension 4: Recency

When was the last meaningful interaction?

Score 0 to 100. 100 means activity within the past 7 days. 50 means activity within the past 30 days. 10 means activity within the past 90 days. 0 means nothing in 90 plus days.

Combining the scores

Multiply the four dimensions. Then divide by 10,000 to keep the number between 0 and 100.

Total score = (Fit × Intent × Engagement × Recency) / 10,000

Multiplying matters. Most legacy scoring models add the dimensions. Addition rewards leads that are great on one dimension and weak on others. Multiplication penalizes them. A lead that is perfect ICP fit but has zero recent activity scores zero, which is the right answer.

Score thresholds

Three tiers.

• MQL threshold: combined score above 25. Lead gets entered into the nurture sequence and tagged for marketing follow-up.
• SQL threshold: combined score above 50. Lead gets routed to a sales development rep for active outreach.
• Sales-accepted threshold: combined score above 70, plus a positive SDR conversation. Lead becomes an opportunity.

Tune the thresholds based on actual conversion rates after the first quarter of data. The starting points above hold for most B2B SaaS businesses.

Where most models break down

Point inflation. Every new event gets assigned 10 points by someone who wants their channel to look more important. By month nine the high scores are unmeaningful.

Fix: lock the maximum point values per dimension. New events have to fit inside the existing 100-point scale. Promote them only if they replace something else.

Negative scoring. Subtracting points for bad signals sounds good and produces messy data. Most negative scoring rules I see do more harm than the leads they are meant to filter out.

Fix: use disqualifiers in the ICP worksheet rather than negative points. Disqualified leads do not enter the scoring system in the first place.

Different scoring rules for different sources. Web leads get scored differently from inbound demo requests, which get scored differently from outbound responses. The complexity multiplies. The trust collapses.

Fix: one scoring model. Lead source is data about the lead, not a modifier to the scoring math.

How to tune the model

Run the model. Wait one quarter. Pull two cohorts: leads that converted to opportunities and leads that did not.

Look at the score distribution for each cohort. The converted leads should cluster above your SQL threshold. The non-converted leads should cluster below. If the distributions overlap heavily, the thresholds need to move or one of the dimensions is mis-weighted.

Adjust threshold values, not point assignments. Stability in the point system is what makes sales trust the score.

Sales adoption

The scoring model only works if sales uses it. Most do not because they were never asked to help build it.

Build the model with at least one senior sales rep in the room. They will push back on the engagement dimension (usually wanting it weighted lower) and the intent dimension (wanting it weighted higher). Both pushes are usually right.

Once the model launches, share the score breakdown with the SDR who works the lead. Not just the combined number. Show them which dimensions scored high and which did not. The transparency builds trust.

Implementation in marketing automation tools

HubSpot, ActiveCampaign, Customer.io, and Marketo all support multi-dimensional scoring. The implementation varies.

HubSpot: Score Properties with separate scores per dimension, then a calculated property for the combined score. Works cleanly.

ActiveCampaign: Lead Scoring with separate scores per category. The combined math runs in an automation.

Customer.io: native attribute-based scoring built around events. The most flexible implementation for product-led teams.

The marketing automation tools comparison covers the platforms in context.

Frequently asked questions

Should I score accounts or contacts?

Both. Score the contact for individual signal. Roll up to an account-level score for buying committee context. Sales acts on the account score. Marketing acts on the contact score.

How often should I tune the model?

Once a quarter, no more. Constant tuning erodes trust. Stability matters more than perfection.

What if my product is PLG and the buying decision happens in-product?

Add a fifth dimension for product engagement. Replace some of the engagement dimension weight with in-product behavior. The math stays the same. Customer.io and HubSpot both handle this cleanly.

Which dimension on your current scoring model is overweighted and you have not been willing to admit it yet?