A HubSpot lead scoring model that sales will trust scores fit and engagement separately, caps how many points any one behavior can add, lets old activity decay, and is tested against real closed-won and closed-lost records before it goes live. HubSpot's current lead scoring tool supports all of that natively. The work is in choosing the criteria, not in clicking the settings.
Most B2B SaaS teams already have a score. The problem is that sales ignores it. Usually that is because one number is trying to say two things at once: "this company looks like our customer" and "this person is active right now." A student who downloads five ebooks can outscore a VP at a perfect-fit account who visited pricing once. Reps see that a few times and stop looking at the score.
This guide covers building the scoring model itself. For the definitions, SLA and routing that sit around it, see our post on the MQL to SQL handoff.
How HubSpot lead scoring works now
HubSpot moved scoring into a dedicated lead scoring tool, found under Marketing > Lead Scoring. It lets you build three kinds of scores:
- Fit scores, based on property values such as job title, company size or industry
- Engagement scores, based on actions such as page views, form submissions, email activity and meetings
- Combined scores, which track fit and engagement together
Scores can be built for contacts, companies and deals, depending on your subscription. HubSpot's documentation lists contact scores under Marketing Hub, company scores under Marketing Hub or Sales Hub, and deal scores under Sales Hub, with Professional or Enterprise required. Check the current requirements in HubSpot's knowledge base before you plan, because they change.
If you still rely on the older HubSpot score property, note that HubSpot sunset legacy scoring in 2025. HubSpot staff said legacy score properties stopped updating after August 31, 2025, and that existing scores were not moved over automatically. If your workflows, lists or reports still reference the old property, they are running on a frozen number.
Step 1: Pull the closed-won and closed-lost data first
Before you open the scoring tool, export the last 12 to 24 months of closed-won and closed-lost deals with their associated contacts and companies. Look for:
- Which industries, company sizes and regions actually close
- Which job titles or departments appear as the champion and the economic buyer
- Which actions happened in the weeks before a demo request or opportunity: pricing page visits, specific feature pages, webinar attendance, trial sign-ups
- Which actions show up just as often on lost or never-converted records
That last list matters most. If an action is common among people who never buy, it should carry little or no weight, no matter how much marketing likes it.
Step 2: Build the fit score from your ICP
Fit is about the account and the person, not their activity. In HubSpot, fit criteria are property-based, organized into groups. Typical B2B SaaS fit groups:
- Company: industry, employee count, revenue band, region, tech stack if you track it
- Person: seniority, department, job function
- Disqualifiers: competitors, students, personal email domains, regions you do not sell to
Give negative points to disqualifiers so a bad-fit record cannot climb into sales' queue on activity alone. HubSpot lets you set limits on each group, so one strong property, like company size, cannot carry the entire fit score.
Step 3: Build the engagement score from buying signals
Engagement criteria are event-based. Weight the actions that preceded real opportunities in your data, not the ones that are easiest to track. For many SaaS teams that means:
- High intent: demo or contact-sales requests, pricing page views, trial sign-ups, product-qualified signals synced from your app
- Medium intent: webinar attendance, case study views, integration or security page views
- Low intent: blog views, newsletter opens, top-of-funnel content downloads
Two settings keep engagement honest:
- Group limits. Cap each group so 40 newsletter opens cannot outweigh one pricing page visit. HubSpot's own example caps a marketing engagement group and a sales engagement group at different maximums.
- Score decay. HubSpot lets engagement points decay over a set period, so activity from six months ago does not keep someone looking hot. HubSpot's example uses 50 percent decay every three months.
Step 4: Combine fit and engagement without blurring them
A combined score in HubSpot creates separate properties for fit, engagement and a threshold label. HubSpot's documentation describes the threshold as a grid: letters A to C for fit and numbers 1 to 3 for engagement, so a record might be A1 (high fit, high engagement) or C1 (low fit, high engagement).
That grid is what earns sales' trust, because it shows why a record scored the way it did:
- A1 and A2: route to sales now
- A3: high fit, quiet. Good for account-based outreach, not a hot lead
- B1 and B2: review, nurture or route by territory
- C1: busy but poor fit. Usually nurture, not sales
Agree on what each cell means in writing with sales leadership before launch. That agreement is part of your MQL definition.
Step 5: Test before you turn it on
HubSpot lets you test a score against individual records and preview how scores will spread across your database before you turn the score on. Use both:
- Score 10 to 20 recent closed-won contacts and 10 to 20 closed-lost or disqualified contacts. Winners should land in A1 or A2. If they do not, the model is wrong.
- Check the distribution. If half your database lands in the top tier, your thresholds are too loose.
HubSpot evaluates scores retroactively when you turn them on, so a bad model will immediately flood or starve the sales queue. Testing first avoids that.
Step 6: Route, then tune monthly
Once live, use a workflow to route records that hit your top threshold to the right rep or queue. HubSpot's example routes high-scoring contacts with a workflow and round-robin rotation. Then review the model every month for the first quarter:
- What share of A1 and A2 records did sales accept?
- What rejection reasons came back, and do they point to a fit property or an engagement action?
- Did any closed-won deals start below threshold? What did they do that the model missed?
Adjust one or two criteria at a time so you can see what changed.
If you want AI scoring
HubSpot also offers AI lead scores: your contact data trains the AI model, which generates recommended criteria and points. HubSpot lists AI contact scoring under Marketing Hub Enterprise. AI scores can be a good cross-check against a manual model, but sales still needs to understand why a record scored high. Keep the fit and engagement logic visible either way.
Make the score part of the system
Lead scoring is one piece of a revenue engine: data hygiene, scoring, routing, SLAs and reporting all have to agree. Our HubSpot revenue operations work configures HubSpot to match how your team actually sells, with proper lead scoring, clean handoffs and reporting that shows what is driving revenue. Book a GTM assessment and we will audit your current HubSpot setup, identify what is broken and show you how to fix it.
Frequently Asked Questions
Where is lead scoring in HubSpot now?
In HubSpot's lead scoring tool, under Marketing > Lead Scoring. It replaced the legacy HubSpot score property, which stopped updating after August 31, 2025, according to HubSpot.
Should fit and engagement be one score or two?
Track them separately, then combine them. HubSpot's combined score keeps fit and engagement as separate properties plus a threshold label like A1 or C1, which makes it clear why a record qualified.
How do I stop old activity from inflating scores?
Turn on score decay for engagement criteria. HubSpot lets points decay by a percentage over a set period.
Which HubSpot subscription do I need for lead scoring?
It depends on the object. HubSpot lists contact scores under Marketing Hub, company scores under Marketing or Sales Hub, and deal scores under Sales Hub, at Professional or Enterprise. Confirm current requirements in HubSpot's knowledge base.
How often should we change the scoring model?
Review it monthly for the first quarter, then quarterly. Change one or two criteria at a time and check the effect on sales acceptance.