The Complete HubSpot Lead Scoring Setup Guide: Turn Cold Prospects Into Qualified Leads

Build a scoring model your sales team trusts, on HubSpot's lead scoring tool.

Most lead scoring does not fail because the maths is wrong. It fails because it scores the wrong signals, nobody revisits it, and sales quietly stops trusting the number. Within a quarter the "hot" list is full of newsletter subscribers while your best-fit buyers sit in the nurture pool.

If you run HubSpot, there is a second problem. The old "HubSpot Score" property stopped updating on 31 August 2025. Any workflow, list or report still pointing at it is working off frozen data.

Here is how to rebuild scoring on HubSpot's Lead Scoring tool so the output drives sales action, not another dashboard nobody opens.

Where lead scoring usually goes wrong

The failure patterns worth naming before you build:

One number for two different questions. Fit (how well someone matches your ICP) and engagement (how interested they are right now) answer different things. Blend them into a single score and a great-fit prospect who has gone quiet looks identical to an active lead you would never sell to. A contact at a competitor browsing your site heavily scores high on activity while being a guaranteed no.

Treating a job title as intent. Seniority and company size matter for fit. They tell you nothing about whether someone is in-market this week. Behaviour does.

A threshold so low everything clears it. If a newsletter signup and a demo request land in the same band, the score is decoration.

Set once, never revisited. Buying behaviour shifts, your ICP sharpens, the product moves on. A model built a year ago is scoring a company you have outgrown.

No negative signals, no decay. Points only ever go up, so a contact who grabbed one guide eight months ago still reads as warm.

What changed in HubSpot

The Lead Scoring tool (Marketing Hub Professional and Enterprise) is more flexible than the legacy property in a few ways worth understanding first:

• Separate Fit and Engagement scores. Score them independently or combine them. Contacts and companies work as Fit, Engagement or Combined. Deals are always combined.
• Scoring across objects. Contacts, companies and deals, not contacts alone.
• Time frames and decay. A time frame scores an action inside a recent window (last 7, 14 or 30 days), after which the points fall away. Decay reduces a whole group of points by a percentage over time, from a one-month minimum.
• Score limits. Cap how many points a repeated action contributes, so one keen contact refreshing your pricing page does not run off with the score.
• Associations and aggregation. Score a contact on their associated company's industry, headcount or revenue. Aggregation set to Minimum or Average stops a company linked to twenty contacts inflating its own score.
• Thresholds and grades. Sort scores into A/B/C and 1/2/3 bands you agree with sales.

One ceiling to plan around: the total tops out at 100 points on Marketing Hub Professional and 500 on Enterprise. Build your bands to fit your tier.

Step 1: Define fit from your closed-won, not your wish list

Pull your recent closed-won deals and look for what they share. Not the logos you want. The ones you win and keep.

• Firmographics: industry, headcount, revenue band, region
• The buyer: role, seniority, who championed it internally
• The pattern: how they found you, what they engaged with before the first call, time from first touch to close

This becomes your fit criteria. If you struggle to describe your best customer in a sentence, the model will guess for you.

Step 2: Build Fit and Engagement separately

Go to Marketing > Lead Scoring and click Create Score. Set up a Contact Fit Score and a Contact Engagement Score as two scores, not one merged number. (Create score with AI gives you a starting point if you want one.)

Keeping them apart is the whole point. Sales sees "great fit, low engagement" (worth a proactive nudge) and "high engagement, poor fit" (worth a polite no) at a glance.

Step 3: Set your fit criteria

Score fit on contact and company properties: role, seniority, industry, headcount, revenue.

• For anything beyond a single property, build a segment (the renamed active list) and score on "belongs to all, any or none of these lists". The tool adds criteria additively and does not chain AND/OR logic inside one rule, so lists carry the complex conditions.
• Use Add object > Associated company to score a contact on their company's firmographics.
• Where a company links to many contacts, set Aggregation to Minimum or Average so the company score stays honest.

Step 4: Set your engagement criteria

Score real buying signals over vanity actions:

• High intent: pricing page views, demo or contact requests, repeat visits in a short window
• Medium: form submissions, content downloads, email clicks
• Low: opens, single page views

Then apply the two features keeping engagement current:

• Time frame for short-lived intent. A pricing page visit in the last 14 days means something. The same visit six months ago does not.
• Decay at group level for the slower signals, for example reducing engagement points by 25% a month from a one-month minimum.

Set Score Limits so one repeated action does not dominate the result.

Step 5: Score the negatives

Points should come off as well as on. Build these as criteria or lists:

• Competitor email domains
• Unsubscribes and hard bounces
• Roles or company sizes outside your market
• For engagement, reset to zero on closed-lost through a workflow, so a dead deal stops reading as warm

Step 6: Set thresholds with sales, not for them

Sort the scores into bands and agree with sales what each one triggers. Grades (A/B/C) for fit and numbers (1/2/3) for engagement pair well. An A1 is a best-fit, actively engaged lead. A C3 is an engaged poor-fit you usually decline.

Write down what each band means in practice:

• Best fit, active: immediate sales follow-up
• Best fit, quiet: proactive outreach or a tailored sequence
• Poor fit, active: low priority, marketing nurture
• Poor fit, quiet: out

If sales does not own the bands, they will not work them.

Step 7: Make the score do something

A score sitting in a property changes nothing. Wire it into the day-to-day:

• Routing and alerts when a contact crosses a threshold, to the right rep, same day
• Deal stage and lifecycle stage updates
• Reset rules for closed-lost and re-engagement
• Reporting on band-to-opportunity and opportunity-to-won, so you learn whether the model predicts revenue or only activity

Review the bands against real outcomes each quarter. If your A1s are not converting better than your B2s, the criteria are wrong, not the leads.

Where to start

Do not build all of this in week one. A workable order:

1. Agree fit with sales and leadership first. It is the hardest part and everything hangs off it.
2. Build the fit score and check it against known good and bad customers.
3. Add engagement, with time frame and decay.
4. Set the bands and the workflows they trigger.
5. Run it for a few weeks, review the spread, adjust.

Scoring is a model you tune as you learn what your best customers do, not a one-off build.

The point of all this

Lead scoring earns its place when it tells a small sales team where to spend its hours. Fit tells you who is worth winning. Engagement tells you who is ready now. Keep them separate and current, and make sure a high score triggers an action rather than a quarterly argument about lead quality.

If your scoring is still running on the retired property, or you have rebuilt it and sales still does not trust the output, the model and the process underneath it have usually drifted apart. Book a call with ROC and we will show you where, and what to fix first.