Table of Content
What is lead scoring?
Lead scoring ranks your leads with a number, so sales talks to the most likely buyers first. Leads earn points for who they are, such as role and company size, and for what they do, such as visiting your pricing page. A lead goes to sales when its score crosses a set threshold.
The short version
- A score is two things. Fit (who they are) and engagement (what they do). Good models use both.
- Start with rules, not machine learning. A simple points model beats a predictive one when you have little closed-deal data.
- Add negative points. Competitors, students, unsubscribes and long silence should lower a score.
- Predictive scoring needs history. HubSpot, Salesforce and Zoho all offer AI scoring that learns from past conversions, and they need enough data to work.
- Review it every quarter. Compare scores with real outcomes and adjust the points.
- Checked on October 15, 2026. Plans and tiers change, so confirm with each vendor.
How a lead score is calculated
Lead scoring assigns points to attributes and actions, adds them up and compares the total with a threshold. A typical setup gives points for fit, such as a decision-maker title, adds points for engagement, such as a demo request, subtracts points for bad signs, and lowers old scores as they age.
Here is an illustrative scoring sheet. The point values are examples, not a standard, so base yours on the traits of customers you have actually won.
| Signal | Type | Points |
|---|---|---|
| Job title is director or above | Fit | +15 |
| Company size in your target range | Fit | +10 |
| Requests a demo | Engagement | +30 |
| Visits the pricing page twice in a week | Engagement | +15 |
| Opens a marketing email | Engagement | +2 |
| Personal email address | Negative | -10 |
| No activity for 60 days | Decay | -20 |
| Works for a competitor | Negative | -50 |
A threshold, say 60 points, marks a lead as ready for sales. Many teams also separate marketing-qualified leads, who show interest, from sales-qualified leads, who match your buyer and have a need.
9 lead scoring models to automate
Each model below can run on its own or be combined. Start with the first three.
1. Demographic fit scoring
Score the person: job title, seniority, department and location. A purchasing director scores higher than an intern. This model never changes with behavior, so it is a stable base.
Choose it if you sell to specific roles. Watch for stale data. Re-check titles on key accounts.
2. Firmographic fit scoring
Score the company: industry, size, revenue, technology used and region. Build it from your ideal customer profile, which is the description of the companies that buy, stay and grow. Data enrichment tools can fill these fields from a company domain.
Choose it if your best customers share traits. Watch for a profile based on opinion. Build it from your closed-won deals.
3. Behavioral engagement scoring
Score actions: demo requests, pricing page visits, content downloads, webinar attendance, email clicks and replies. Weight high-intent actions far above casual ones. A demo request is worth more than twenty email opens.
Choose it if you have web and email tracking. Watch for noise. Do not let low-value actions add up to a false signal.
4. Intent signal scoring
Add outside signals that show a company is researching your category, such as visits to review sites or searches for related topics, where your tools provide them. Treat intent data as a hint and confirm with a real action.
Choose it if your sales cycle starts well before a form fill. Watch for cost and false positives. Test on a small segment first.
5. Negative scoring
Subtract points for signs a lead will not buy: personal email domains for a B2B product, students, competitors, unsubscribes, bounced emails and job-seeker visits to your careers page. This keeps junk out of the sales queue.
Choose it if your sales team complains about lead quality. Watch for over-filtering. Review rejected leads each month.
6. Time-decay scoring
Interest fades. Reduce the value of old actions or subtract points after a quiet period, so a demo request from six months ago does not outrank a pricing visit yesterday. A simple rule is to lose a set number of points every 30 days without activity.
Choose it if your pipeline holds many stale leads. Watch for cutting too fast. Match decay to your sales cycle length.
7. Predictive (AI) scoring
The model learns from past conversions and scores new leads by their likelihood to buy. HubSpot offers predictive scoring that produces a likelihood-to-close property, though sources disagree on which subscription tier includes it, so check HubSpot’s current plan page. Salesforce Einstein Lead Scoring shows the top positive and negative factors for each lead, starts from a global model and shifts to a personalized one when enough of your data exists, and refreshes its models about every 10 days. Zoho’s Zia Scores works in a similar way and needs a minimum amount of data.
Choose it if you have a large history of won and lost deals. Watch for black boxes. Look at the factors the tool shows, and compare scores with outcomes.
8. Account-based scoring
Score the whole company, not one person. If three people from the same account visit your site this week, the account score rises. This fits B2B deals with several decision-makers.
Choose it if your deals involve buying committees. Watch for missing links between contacts. Clean up account matching first.
9. A fit-and-engagement grid
Instead of one number, plot leads on a grid: fit (A, B, C) and engagement (1, 2, 3). An A1 lead goes to sales now, an A3 goes into nurture and a C1 gets a light-touch reply. The grid is easy to explain and hard to game.
Choose it if you want sales to trust the system. Watch for too many cells. Keep three levels on each axis.
Lead scoring software, including HubSpot lead scoring
Most CRMs include scoring. Check what you already own before you buy a separate tool.
| Tool | Rule-based scoring | AI scoring | Notes |
|---|---|---|---|
| HubSpot | Manual scoring, reported on Professional tiers and above | Predictive likelihood-to-close; tier reported differently by sources | Verify tiers on HubSpot’s pricing page |
| Salesforce | Scoring through rules and Account Engagement | Einstein Lead Scoring with top positive and negative factors | Trailhead lists Sales Cloud Einstein or High Velocity Sales licenses |
| Zoho CRM | Scoring rules under Automation | Zia Scores, separate from manual rules | Needs a minimum amount of data |
| Spreadsheet | Points sheet you update | None | Good for a first model on a small list |
If your leads come from several sources, a customer data platform can unify them before scoring. Our guide to customer data platforms explains how, and our CRM software comparison helps if you need a new system.
How to build a lead scoring model in 7 steps
- Look at your closed-won deals. List the titles, company traits and actions they shared before they bought.
- Define the fit score. Use those traits to set demographic and firmographic points.
- Define the engagement score. Weight the actions that came before purchases the most.
- Add negative and decay rules. Remove junk and age out old interest.
- Set a threshold and a handoff. Decide the score at which sales is alerted, and what sales must do within what time.
- Review quarterly. Compare scores with outcomes. If high-scoring leads do not close, change the points.
- Match the follow-up to the score. High scores get a call, middle scores get nurture emails. Our guides to email marketing strategy for startups and AI marketing cover the nurture side.
Where AI adds value, and where it does not
AI lead scoring helps in three ways: learning which traits predict a sale from your history, enriching records with missing company data, and writing a short reason a lead was scored high so sales trusts it. It does not fix a weak process. If sales never follows up, no score will help, and if you have only a few closed deals, a predictive model has little to learn from.
For the sales side of the same workflow, read our guide to the AI sales agent, which covers agents that qualify and follow up with leads. Our guide to the best AI agents for business compares agent options, and how to build an AI agent for your business shows the steps.
When to build a custom lead scoring workflow
CRM scoring covers common cases. A custom workflow helps when scores need data from your product, billing, support and ad platforms, or when you want an agent to score, enrich, route and book a call in one flow. Our AI automation services team builds lead qualification agents, and our AI workflow automation service connects them to your CRM and calendar.
Frequently Asked Questions
What is lead scoring?
Lead scoring is ranking leads with points for fit and engagement, so sales contacts the most likely buyers first. A lead is passed to sales when its score crosses a threshold.
How is a lead score calculated?
Add points for fit, such as job title and company size, and for actions, such as a demo request or pricing page visit. Subtract points for negative signs and reduce old scores over time. Compare the total with a threshold.
What are examples of lead scoring criteria?
Fit criteria include job title, company size, industry and region. Engagement criteria include demo requests, pricing page visits, content downloads and email replies. Negative criteria include personal email addresses and competitors.
What is the difference between lead scoring and lead qualification?
Lead scoring ranks leads with numbers. Lead qualification checks whether a lead has the need, budget and authority to buy. Scoring decides who to look at first, and qualification decides who is a real opportunity.
What is predictive lead scoring?
Predictive lead scoring uses a model trained on your past conversions to estimate each lead’s chance to buy. HubSpot, Salesforce Einstein and Zoho Zia Scores offer versions, and all need enough historical data to be useful.
Does HubSpot have lead scoring?
Yes. HubSpot offers manual scoring and predictive scoring, though the tiers that include each are reported differently by sources, so check HubSpot’s current pricing page.
Does Salesforce have lead scoring?
Yes. Salesforce offers Einstein Lead Scoring, which shows the top positive and negative factors for each lead and moves from a global to a personalized model as your data grows.
How many points should a lead need to go to sales?
There is no standard. Pick a threshold from your own data, such as the score at which past leads converted at a healthy rate, and adjust it quarterly.
How often should I review a lead scoring model?
Review it every quarter. Compare scores with actual outcomes and ask sales which high-scoring leads were poor and which low-scoring leads closed, then adjust points.
Bottom Line
The best lead scoring models are simple, built from your own closed deals and checked against outcomes. Start with fit and engagement, add negative and decay rules, and move to predictive scoring only when you have the history for it.
When you want scoring, enrichment and routing to run as one workflow, our AI automation services team can scope a lead qualification agent with you.
