How to Set Up Lead Scoring in Your CRM (A Simple Points Model)

Most sales teams have the same complaint about marketing leads: “Half of these people were never going to buy.” And most marketing teams have the opposite complaint: “Sales never follows up on the leads we send.”

Lead scoring fixes both, when it is done well. It gives every lead a number based on how well they match your ideal customer and how interested they seem. High scores go to sales fast. Low scores stay in nurture. Everyone works from the same rules.

The trouble is that many lead scoring models are so complicated nobody trusts them. This guide shows you a simple points model built on 2 types of signals, fit and behavior, that you can set up in almost any CRM in an afternoon and improve over time.

What lead scoring is and why it matters

Lead scoring is a system for ranking leads by how likely they are to become customers. Each lead gets points for attributes and actions that signal a good prospect, and sometimes loses points for signals that suggest a poor fit.

Done well, lead scoring:

• Speeds up follow-up on the best leads, when interest is highest.
• Saves sales time by keeping low-quality leads out of their queue.
• Creates a shared definition of a qualified lead between marketing and sales.
• Shows marketing which campaigns bring in leads that actually turn into deals.

You need a CRM with your leads and their activity in one place before scoring is useful. If you are still choosing one, see our guide on what a CRM is.

The 2 types of signals: fit and behavior

Every good scoring model separates 2 questions.

Fit: Is this the kind of customer we want? Fit is based on who the lead is. For a business-to-business company that usually means industry, company size, job title, location and budget. For a consumer or local business it might be location, project type or timeline.

Behavior: How interested are they right now? Behavior is based on what the lead does: visiting your pricing page, requesting a demo, opening and clicking emails, attending a webinar, replying to a salesperson.

Keeping them separate matters. A perfect-fit lead who has done nothing yet needs a different approach than a poor-fit lead who downloads everything you publish. Several CRMs reflect this split. HubSpot’s lead scoring tool, for example, offers fit scores based on property values, engagement scores based on actions, and combined scores that show both.

How to build a lead scoring model step by step

Follow these 7 steps in order. Steps 1 to 6 get you live; step 7 keeps the model useful.

Step 1: Get sales and marketing in the same room

Before you assign a single point, hold a 1-hour meeting with sales and marketing. Bring a list of your last 20 to 50 closed-won deals and a similar number of lost or unqualified leads. Ask:

• What did our best customers have in common before they bought?
• Which actions did they take before talking to sales?
• Which leads waste the most time, and what do they have in common?
• What is the minimum information sales needs before calling?

Write down the answers. This is the most important step, because sales will only trust a model they helped build.

Step 2: Build your fit score

List 4 to 6 fit attributes and assign points. Keep the numbers simple and give the biggest points to the attributes that best predicted closed deals in your review.

Here is an example for a business software company selling to US companies with 20 to 500 employees:

Fit attribute	Value	Points
Company size	20 to 500 employees	+20
Company size	Under 5 employees	-10
Industry	Target industries (e.g., healthcare, professional services)	+15
Job title	Owner, director, VP or C-level	+15
Job title	Student or job seeker	-30
Location	Countries or states you serve	+10
Location	Regions you do not serve	-20
Email domain	Business email	+5
Email domain	Free personal email (only if this matters in your market)	-5

Negative scores are just as important as positive ones. They stop students, competitors and out-of-area leads from piling up high behavior scores and landing in a salesperson’s queue.

For local and consumer businesses, swap in your own attributes. A roofing company might score zip code, project type (repair vs full replacement) and timeline. A real estate agent might score price range, pre-approval status and move date; our guide to choosing a CRM for real estate agents covers that use case.

Step 3: Build your behavior score

Now list the actions that signal buying interest and assign points based on intent. High-intent actions, such as requesting a demo or a quote, should be worth far more than passive ones, such as opening an email.

Behavior	Points	Why
Requested a demo, quote or consultation	+40	Direct buying intent
Visited the pricing page	+15	Evaluating cost
Visited the pricing page 3 or more times in 30 days	+15 more	Strong active interest
Replied to a sales email	+20	Engaged with your team
Attended a webinar or event	+10	Invested time
Downloaded a guide or checklist	+5	Early research
Clicked a marketing email	+3	Light engagement
Opened a marketing email	0	Too unreliable to count
Visited the careers page	-10	Likely a job seeker
Unsubscribed from email	-20	Lost interest

A note on email opens: privacy features such as Apple’s Mail Privacy Protection prevent senders from seeing whether a message was opened, which makes open data unreliable as a scoring signal. Use clicks and replies instead.

Step 4: Add score decay

Interest fades. Someone who visited your pricing page 8 months ago is not the same as someone who visited yesterday. Score decay reduces behavior points as they age.

Many CRMs support this directly. HubSpot’s scoring tool, for instance, can reduce engagement points on a schedule of 1, 3, 6 or 12 months. If your CRM doesn’t support decay, you can approximate it by only counting actions from the last 90 days, or by running a workflow that resets behavior scores after a period of inactivity.

Don’t decay fit scores. A company’s size and industry don’t change because they went quiet.

Key takeaway: Keep fit and behavior separate, use negative points to filter out bad leads, and let behavior points decay. A simple model that sales trusts beats a complex one nobody uses.

Step 5: Set thresholds and the handoff process

With both scores in place, decide what happens at each level. A simple grid works well:

	Low behavior (under 20)	Medium behavior (20 to 49)	High behavior (50+)
High fit (40+)	Targeted nurture or sales prospecting	Sales follow-up within 2 business days	Sales follow-up the same day
Medium fit (15 to 39)	General nurture	Nurture, plus sales review weekly	Sales follow-up within 1 business day
Low fit (under 15)	No sales outreach	Nurture only	Quick sales check, then disqualify or nurture

Leads that cross your threshold are often called marketing qualified leads (MQLs). Once sales accepts and works them, they become sales qualified leads (SQLs). Agree in writing on:

• The exact score that triggers handoff.
• How fast sales follows up.
• How sales marks a lead as accepted, rejected or recycled, and why.

Then automate the handoff: a CRM workflow that assigns the lead to a rep, creates a task and sends an alert when the threshold is crossed. Our guides to building a sales pipeline and email automation workflows show how to connect these steps.

Step 6: Test against past deals before you launch

Before turning the model on, score a sample of past leads, both won and lost, using your new rules. Ask:

• Did most closed-won deals score above the threshold?
• Did most bad leads score below it?
• Are there obvious misses, such as great customers who would have scored low?

Adjust points until the model separates good and bad leads reasonably well. It will never be perfect. The goal is better than today, not flawless.

Step 7: Review every quarter

Lead scoring is not set-and-forget. Every quarter, pull a report and check:

• Conversion by score band. Do higher scores actually convert to deals at higher rates?
• Sales rejection reasons. If sales rejects many MQLs for the same reason, add a negative score for it.
• New behaviors. New content, events or product pages may deserve points.
• Threshold volume. If sales is overwhelmed, raise the threshold. If they are starved, lower it or improve lead generation.

Change 1 or 2 things at a time so you can see what made the difference.

Rules-based vs predictive lead scoring

Many CRMs now offer AI-driven or predictive lead scoring alongside manual points models. Predictive scoring looks at your historical leads and conversions and finds patterns on its own.

	Rules-based (points)	Predictive (AI)
How it works	You choose attributes and points	Software learns from past conversions
Data needed	Works from day 1	Needs a meaningful history of leads and conversions
Transparency	Easy to explain to sales	Can be harder to explain
Maintenance	Manual quarterly reviews	Retrains automatically
Best for	Smaller teams and newer businesses	Teams with large lead volumes and clean data

Predictive tools have real data requirements. Salesforce’s Einstein Lead Scoring setup considerations, for example, call for at least 1,000 leads created in the last 200 days, with at least 120 of them converted, to build a model from your own data; otherwise it falls back to a global model built on anonymized data from many customers. Smaller businesses usually get more value from a clear points model first. Our explainer on predictive analytics covers how these models work.

Common mistakes to avoid

• Too many rules. 30 scoring criteria are hard to explain and harder to maintain. Start with 10 to 15.
• Scoring email opens heavily. They are unreliable. Use clicks, replies and website actions.
• No negative scoring. Job seekers and students end up as “hot leads”.
• No decay. Old activity keeps cold leads looking warm forever.
• Skipping sales input. A model built only by marketing rarely gets adopted.
• Messy data. Duplicate contacts and empty fields wreck fit scores. Our CRM implementation checklist covers data cleanup.

The bottom line

Lead scoring works when it is simple, shared and maintained. Score fit and behavior separately, use negative points and decay, set a clear handoff threshold with agreed follow-up times, and check the results against real deals every quarter. Start with a points model your sales team helped design, and consider predictive scoring later once you have the data volume to support it.

Sources

• HubSpot Knowledge Base: Overview of the lead scoring tool — fit, engagement and combined scores
• HubSpot Knowledge Base: Build lead scores — score decay intervals, score limits and event groups
• Apple Support: Use Mail Privacy Protection on iPhone — why email open data is unreliable for scoring
• Salesforce Help: Considerations for setting up Einstein Lead Scoring — minimum lead and conversion data for predictive scoring and global model fallback