HubSpot Lead Scoring: Engagement, Fit and Combined Scores, Limits and Decay

Here's what trips people up. Contact scores are Marketing Hub only and deal scores are Sales Hub only, so a Sales Hub account on its own can score companies and deals, but not contacts. Building a score with AI needs Marketing Hub Enterprise, and the separate predictive score, Likelihood to close, needs Marketing Hub or Sales Hub Enterprise.

Which hub scores which records

Record	Hub you need, as of October 2026	Score types you can build
Contacts	Marketing Hub	Engagement, fit or combined. AI-built engagement or fit scores on Marketing Hub Enterprise only
Companies	Marketing Hub or Sales Hub	Engagement, fit or combined
Deals	Sales Hub	Combined only

View permission for Lead Scoring lets someone see the tool and score values; Edit permission lets them create and change scores. You can build several scores per object, so sales, support and each region can have their own. Label every score property clearly so each team knows which one is theirs.

Engagement, fit and combined: what each one measures

Engagement scores look at actions, such as visiting your website, subscribing to your newsletter, clicking a CTA or opening a marketing email. Fit scores look at property values, such as job title, company size or annual revenue. Combined scores use both and create three properties: total, engagement and fit.

For company and deal scores, events come from associated contacts, and Association settings decides how their points roll up: Sum, Average, Min or Max. Call events only count calls made through HubSpot's own calling tool.

How the points add up

Every score has at least one group, and groups add up to the total. There are three limits. The score limit caps the whole score, with options from minus 100 to 100 up to minus 10,000 to 10,000. A group limit caps what one group can add or take away. Criteria points are what each rule adds or subtracts, and decimals are allowed. Once a record hits a limit, no more points move in or out.

HubSpot's worked example: the overall limit is 100. A sales group is capped at 60, with 15 points for a started call, 10 for a booked meeting and 20 for a completed meeting. A marketing group is capped at 40, with 6 points for a CTA click, 2 for an email open and 5 for an email link click. A contact who books a meeting, opens two emails and clicks three links scores 10 in the sales group and 19 in the marketing group, so 29 overall.

Separate rules score independently, even inside one group, while extra criteria inside one rule use AND logic. The operators behave differently by score type:

Setting	On fit criteria	On engagement criteria
Contains any of	Adds or subtracts points once, however many values match	Adds points for each matching event
Score together (the default)	One set of points when any value matches	The same points each time any of the chosen events happens
Score individually	Each matching value adds its own points, then they're added together	Each event adds its own points, then they're added together

You can subtract as well as add, say for a newsletter unsubscribe, and a score goes negative if more comes off than goes on. By default an event scores every time it happens. Switch Score every time to Limit to and you cap it: a form worth one point submitted five times, limited to two points, only counts twice.

Score decay: why old clicks stop counting

On engagement or combined event groups you can turn on Decay scores. It cuts an event's points based on how long ago it happened, at intervals of every 1, 3, 6 or 12 months, with a month counted as 30 days. It's worked out from the original value, so a form fill worth 10 points with half decaying each month is worth 5 after a month and nothing 30 days after that.

Decay applies to history too. If a CTA click is worth 2 points and fully decays in three months, a click from four months ago adds nothing when you turn the score on. It doesn't work on segment membership criteria, because membership is a yes or no with no timestamp. The workaround is a date filter inside the segment, such as contacts who submitted a form in the past 30 days.

Thresholds that turn numbers into labels

Thresholds create an extra property with colour-coded labels. Engagement and fit scores use High, Medium and Low, for example 70 to 100 as High and 40 to 69 as Medium. Combined scores use A1 to C3. The letter is fit, from A for high fit to C for low fit, and the number is engagement, from 1 for high to 3 for low, so a C1 is a poor fit who's very engaged.

Building your first score

1. Click More, then go to Marketing, then Lead Scoring. If More isn't there, go to Marketing, then Lead Scoring directly. Click Create score.
2. Choose Contacts, Companies or Deals.
3. Pick a fit, engagement or combined score. Deal scores are combined by default. Click Create.
4. Click Add event group, Add property group or Discover AI rules, then set the score limit, group limits and points for each rule.
5. On the records tab, score everything and exclude segments, or score only members of chosen segments.
6. On the Settings tab, edit the score properties and set your thresholds.
7. Click Test a record or Preview distribution, then Review and turn on. Leave without turning it on and it's saved as a draft.

When a score goes live, records are scored retroactively from current and historical data. Edit it later and everything is re-evaluated, which can affect views, segments, workflows and reports using the property, so check its Used In tab first.

AI scores and predictive Likelihood to close

AI scores need Marketing Hub Enterprise and work for contact engagement and fit scores. HubSpot trains on your own contacts, and it needs at least 50 of them: 25 that converted and 25 that didn't. Give it a start and end point, such as Marketing Qualified Lead to Sales Qualified Lead over 30 days, and it suggests criteria from what those contacts shared.

Predictive scoring is a different thing. On Enterprise, HubSpot estimates the chance an open contact becomes a customer within 90 days and stores it in Likelihood to close. Contact priority then sorts contacts into Very High, High, Medium and Low, with a quarter of your contacts in each, so the ranges shift over time. You can't edit either property. Inputs include page views, email opens and clicks, logged notes and whether the email is on a free mail domain like Gmail. It's a blackbox model, so you can't see how each input affects a score.

My verdict on HubSpot lead scoring

I'd start with one engagement score built the way HubSpot's own Sprocket example does it: a marketing group capped at 40 for lighter signals like opens and page views, with decay on those rules, and a sales group capped at 60 for demo requests and meetings. Then point a HubSpot workflow at the High threshold so it assigns a rep and sends them a notification. I'd skip AI scores until you've got enough closed history to train them, and if you're choosing an edition, check what each Marketing Hub edition includes first, because without Marketing Hub you can't score contacts at all.

A lead score tells your team who to call next. Getting the right people to find you in the first place, including in AI answers, is the part I help with. I run AI training and workshops for teams, and help businesses with AI enablement, workflow automation and GEO, which is about getting your business seen and cited in AI answers. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.

Frequently asked questions

Is HubSpot lead scoring available on Starter?

No. As of October 2026 the lead scoring tool needs Marketing Hub or Sales Hub Professional or Enterprise.

Can you score contacts with Sales Hub?

No. Contact scores are Marketing Hub only. Sales Hub lets you score companies, and deals with combined scores.

How does score decay work in HubSpot?

It reduces an event's points based on how long ago it happened, at intervals of 1, 3, 6 or 12 months, worked out from the original value. It applies to past events too, and it doesn't work on segment membership criteria.

What do A1 and C3 mean in a HubSpot combined score?

The letter is fit, from A for high fit to C for low fit, and the number is engagement, from 1 for high to 3 for low.

How many contacts do you need for an AI lead score?

At least 50, with 25 that converted and 25 that didn't.