HubSpot Lead Scoring Setup: A Step-by-Step Technical Guide

Your sales team has 400 contacts and no idea which ones to actually call. So they sort by "last activity" and start at the top. Some of those leads are solid. Most aren't. That's the problem hubspot lead scoring setup exists to solve.

Building the system takes about two hours if you've done the thinking first. The thinking is the part that takes a week.

Two Scoring Models, One Decision to Make

Manual scoring (Contact Score). You define the rules. A demo request is worth 20 points. A VP-level job title adds 15. A contact who unsubscribed loses 50. HubSpot recalculates the score in real time as contacts meet or stop meeting those criteria. You can explain every number to a skeptical sales rep.

Predictive scoring. HubSpot's machine learning model analyzes your historical closed-won data and assigns scores based on patterns it identifies. Available on Marketing Hub Enterprise only, which starts around $3,600 per month. It's a black box. You can't explain why a contact scored a 73 without digging into the model's internal logic, which you don't have access to.

For most B2B SMBs, manual scoring is the right starting point. Predictive scoring needs enough historical data to train on (typically 200+ closed deals) to produce anything reliable. And even when it works, you lose the ability to defend a score in a sales/marketing alignment meeting. Start manual. Revisit predictive when you have the data and budget. For full context on what Enterprise costs versus Professional, see our breakdown of what HubSpot implementation actually costs.

Do This Before You Open HubSpot

The scoring system you build is only as good as the criteria you choose. Get this wrong and you'll spend six months wondering why the sales team ignores the score field.

Pull your last 20-30 closed-won deals. Look for patterns. Which pages did those contacts visit before converting? What did they download? What job titles and company sizes appeared most often? What actions did they take in the week before requesting a demo?

Separate fit from intent. Fit criteria are about who a contact is: job title, company size, industry. Intent criteria are about what they've done: visited the pricing page, submitted a demo form, attended a webinar. Both matter. Intent almost always predicts conversion better than fit alone.

Agree on a threshold before you build anything. The MQL threshold is the score at which a contact gets handed to sales. If your team can't align on what a qualified lead looks like before you build the system, the score output won't resolve that disagreement. Get the conversation on the calendar now.

A starting framework:

Criterion Type	Example	Suggested Points
High-intent behavior	Demo request, pricing page (2+ visits)	+15 to +25
Mid-intent behavior	Content download, webinar attendance	+5 to +10
Low-intent behavior	Newsletter signup, single blog visit	+2 to +5
Fit: strong match	Job title contains "VP," "Director," "Head of"	+10 to +20
Fit: partial match	Company size 10-200 employees	+5 to +10
Negative: disqualifier	Unsubscribed, personal email domain	-25 to -50
Negative: weak signal	No engagement in 90 days	-10 to -20

Start with 8-12 criteria total. Adding 40 criteria before you have data on which ones actually predict conversion is how you end up with a scoring system that produces numbers nobody understands or trusts.

HubSpot Lead Scoring Setup: Step by Step

Step 1: Find the scoring tool

Navigate to Marketing > Lead Scoring in the left nav. In some portal configurations, you can also reach it at Settings > Properties, then search for "HubSpot Score." Both routes land in the same place. You'll see a split view: Positive Attributes on the left, Negative Attributes on the right.

Step 2: Add your positive attributes

Click "Add criteria" under Positive. HubSpot lets you score on:

• Contact properties: job title, lifecycle stage, any custom property you've created
• Company properties: industry, employee count, annual revenue
• Form submissions: a specific form, or any form
• Email engagement: opens, clicks (not unsubscribes; that's a negative)
• Page views: by specific URL, or URL contains a keyword like "/pricing"
• List membership: useful if you're already segmenting by ICP fit
• Deal activity: contact is associated with an open deal

Set each criterion's point value using the weights from your pre-work. Keep it to 12 or fewer for now.

Step 3: Add your negative attributes

Same interface, right column. This is the part most teams skip entirely. Common negative criteria worth adding from day one:

• Unsubscribed from email: -50
• Job title contains "student," "intern," or "freelance": -30
• Email domain is personal (gmail, yahoo, hotmail, outlook.com): -25
• Company employee count under 5: -20
• No email opens or page views in the last 90 days: -15

A contact who looks great on paper but hasn't interacted with anything in three months should not be sitting at the top of your sales queue. Negative attributes are what keep the score honest over time.

Step 4: Wire it to a workflow

Lead scoring without a workflow is just a number on a contact record. Build a contact-based workflow with these settings:

• Trigger: HubSpot Score is known AND HubSpot Score is greater than or equal to [your MQL threshold, e.g., 50]
• Re-enrollment: On (fires again each time a contact crosses back above the threshold after falling below it)
• Action 1: Set Lifecycle Stage to Marketing Qualified Lead
• Action 2: Notify the assigned contact owner, or create a follow-up task for sales
• Optional branch: Different actions for score 50-75 versus 75+, if your team wants to differentiate urgency

Step 5: Validate before you ship

Before the system runs on your full database, pull 15-20 contacts you know well. Some you'd consider strong leads. Some weak. Run through their scores manually. Does the output match your intuition? If a contact who closed as a paying customer is sitting at 8 points, something is wrong with your criteria or your weights. Fix it before you let the workflow fire at scale.

When Lead Scoring Makes Things Worse

Short sales cycles. If your average deal closes in under two weeks, behavioral scoring has limited value. The contact doesn't accumulate enough signal before the conversation is already won or lost. Lead scoring pays off most when your cycle is 30+ days and you're managing real inbound volume.

Very small inbound volume. Lead scoring is a prioritization tool. If your team gets 20 new inbound leads per month, they can evaluate each one personally. The setup overhead isn't worth it until you're dealing with volume you can't reasonably eyeball. A rough threshold: it starts making sense around 80-100+ new contacts per month.

No sales/marketing alignment on the MQL definition. If marketing builds the criteria alone and sales doesn't trust the output, reps will ignore the score field and you'll have a system nobody uses. Build the criteria together and agree on the threshold together. Then schedule a monthly review. If none of that is happening, fix the alignment problem first.

Pure outbound-only pipelines. If your leads never interact with your website or content before a conversation, you don't have behavioral data to score. Fit-only scoring is a limited prioritization tool when you're already selecting who to contact. Scoring makes far more sense when there's real inbound activity to rank.

Maintaining What You Built

The criteria you set today will be partially wrong in six months. That's fine.

Check your score distribution monthly. If 60% of contacts are above your MQL threshold, the threshold is too low. If fewer than 5% ever cross it, your criteria are too strict or your inbound quality needs attention. Aim for something in the 10-25% range hitting MQL status over a 90-day window.

Review closed-won and closed-lost deals quarterly. Did your highest-scoring contacts convert at meaningfully higher rates? Which specific criteria showed up most often in contacts who closed? Drop the criteria that aren't predictive. Add new ones based on what you're observing.

One pattern I see regularly: the scoring system stops matching reality when the ICP shifts. A company that moved upmarket six months ago probably has a different ideal lead profile than their scoring rules reflect. Update the criteria when the business changes.

Frequently Asked Questions