How to Set Up Lead Scoring in HubSpot

Before you assign a single point

It's tempting to open HubSpot's scoring tool and start assigning points to properties — job title, page views, email opens — because the interface makes it look like configuration, not design. It's design. Get the design wrong and the score becomes a number nobody on the sales team actually checks.

A real setup is five decisions, made before any point value gets typed in.

Decision 1: separate fit from engagement

A single blended score conflates two different questions: does this account match who you actually sell to (fit), and is this specific person, right now, showing real buying intent (engagement)? Blend them and a perfectly-matched account with zero activity scores the same as a poorly-matched one that just downloaded a whitepaper — both look mediocre, for completely different reasons.

Score fit and engagement as two separate properties. Combining them comes later; scoring them together from the start loses the information you need to combine them well.

A score that can't tell you whether the problem is fit or intent isn't giving a rep anything to act on.

Decision 2: build fit from your actual customers, not your aspirations

Fit scoring works backwards from your existing customer base — the firmographic and demographic traits (company size, industry, role) that show up disproportionately in accounts that actually became good customers, not the traits of the account you'd love to land.

Keep this component short. Four or five criteria that genuinely separate good-fit from poor-fit accounts beats fifteen that mostly agree with each other.

Decision 3: weight engagement by recency, not just volume

Engagement scoring is where most setups quietly become unfair to time: a prospect who visited the pricing page yesterday and one who visited it eight months ago can end up with the same point total, because most models just sum activity over all time.

The fix is a decay function, not a bigger point value — let engagement points fade the longer since the activity happened, so the score reflects current intent instead of a diary of everything that's ever happened.

Decision 4: combine them as a matrix, not a sum

Adding fit and engagement together as one final number throws away exactly the distinction Decision 1 was built to preserve. A high-fit, low-engagement account (worth nurturing) and a low-fit, high-engagement account (worth requalifying, not routing to your best AE) need different next actions — a summed score treats them identically whenever the totals happen to match.

Route by the combination, not the sum: high fit and high engagement moves fast; high fit and low engagement gets nurtured; low fit and high engagement gets requalified before it gets routed at all.

Decision 5: connect the score to an action, or it's just a number

A score that doesn't trigger anything — no routing change, no alert, no lifecycle-stage shift — trains the sales team to ignore it within a month. The model isn't finished until it's wired into whatever happens next: assignment, a lifecycle stage, a rep notification.

This is the same logic behind a real system that cut inbound assignment time from over 80 days to 71 minutes — the full story of what happens once scoring is actually wired to action is in the case study.

What this looks like, built

Five decisions, applied for real: