Lead Generation10 min read

Lead Scoring Model: How Do You Build One That Sales Trusts?

A lead scoring model ranks leads by fit and buying intent so sales works the best ones first. Here is how to build one, set a handoff threshold and keep scores from going stale.

Hamza AltafPublished Nov 15, 2026
Share
Blog cover reading "Lead Scoring Model: How to build one sales trusts" over a photo of two colleagues sorting index cards into ranked piles on an office desk, illustrating a lead scoring model.

A lead scoring model is a points system that ranks leads by how well they fit your ideal customer and how strongly they are showing intent to buy, so sales works the best leads first. Most models fail not because the math is wrong, but because sales stops believing the number. This guide covers what goes into a model, how to set points and a handoff threshold, how to keep scores from going stale, and a simple scoring framework you can copy.

Key takeaways

  • Score two things separately: fit (who the lead is) and intent (what the lead is doing). A single blended number hides which one is weak.
  • Build the first version from your closed-won and closed-lost deals and from interviews with sales, not from guesses about what feels important.
  • Use negative points and score decay. Without them, scores only ever go up and old activity looks like fresh interest.
  • Set the handoff threshold so it takes several meaningful actions to reach it, while a direct buying action like a demo request reaches it on its own.
  • Treat the model as a draft. Review it against real outcomes every quarter and adjust.

What is a lead scoring model?

A lead scoring model is a set of rules that adds or subtracts points from a lead based on who they are and what they do, then turns the total into a priority. It answers one question for a rep: who should I call first?

Two of the most widely used marketing platforms split scoring into the same two halves. HubSpot’s lead scoring tool offers fit scores based on property values, engagement scores based on events, and combined scores that use both. Adobe’s Marketo Engage guidance describes the same split as demographic scoring for fit and behavioral scoring for interest and purchase intent, and recommends combining them to find the best leads on both counts.

Scoring sits after qualification criteria in the workflow. If you have not decided what a qualified lead looks like yet, start with our guide to lead qualification frameworks, then use scoring to apply those criteria at scale.

What’s the difference between fit scoring and engagement scoring?

Fit scoring measures whether a lead could be a good customer. Engagement scoring measures whether they are acting like a buyer right now. You need both, because each one fails on its own.

Fit scoreEngagement (intent) score
Question it answersShould we ever sell to this account?Is this person in a buying moment now?
Typical inputsIndustry, company size, job title, department, regionDemo requests, pricing page visits, content downloads, event attendance, replies
Changes over time?Slowly, when the company or role changesQuickly, and should fade if activity stops
Fails alone becauseA perfect-fit account with no current need wastes a rep’s callA student downloading every ebook looks like a hot lead
Where the criteria come fromYour ideal customer profileActions that preceded past closed-won deals

Your fit criteria should come straight from your ideal customer profile. If the ICP is vague, the fit score will be too.

How do you build a lead scoring model step by step?

Start from evidence you already have: the deals you won, the deals you lost, and what your sales team says separates them. Then turn those patterns into points, test them against history, and launch with a review date.

Adobe’s Marketo guidance lays out a practical order: list the activities that matter to your sales process, ask sales which actions signal a good or bad lead, analyze closed-won and closed-lost opportunities for patterns, and interview sales about which job titles sign off on purchases. It also frames scoring as iterative, with scores and thresholds adjusted as more data comes in.

  1. Pull your last batch of closed-won and closed-lost deals. Note the industry, company size and buyer title on each, plus the actions the lead took before the first sales conversation.
  2. Interview two or three reps. Ask which signals made them take a lead seriously and which ones turned out to be noise.
  3. Write fit rules. Give points for attributes that show up in won deals and subtract points for attributes that show up in lost ones.
  4. Write intent rules. Weight actions by how close they are to a purchase decision, not by how easy they are to count.
  5. Back-test. Score last quarter’s leads with the new rules. If the won deals do not cluster near the top, the weights are wrong.
  6. Launch with a review date. Put a quarterly check on the calendar before the model goes live.

Should you use a rules-based or a predictive lead scoring model?

Rules-based models use points you set by hand. Predictive models learn from your history. Salesforce describes its Einstein Lead Scoring as one that analyzes your history of lead conversions, scores each lead on how closely it matches those patterns, shows which field values most affect each score, and keeps adjusting as more leads convert. Predictive scoring works best when you have a meaningful history of conversions to learn from. If you are early, a transparent rules-based model is easier to explain to sales and easier to fix.

How many points should each action be worth?

Give the most points to actions that sit closest to a purchase, a few points to light research, and negative points to signs of poor fit or lost interest. The exact numbers matter less than the ratios between them.

Adobe’s published example shows the spread well. It assigns +50 to a demo request form, +15 to a trade show visit, +5 to a white paper download and -10 for no interaction in the last 90 days. On the fit side, the same example gives extra points to target industries and company sizes and subtracts points for companies under 10 employees. A demo request is worth ten white papers in that model, which is roughly how a sales rep would rank them too.

The FindyLeads Fit-Intent-Freshness score

Here is a simple starting framework we use when thinking about scoring. It splits the total into three parts so a rep can see why a lead ranks where it does. The point values are a starting template, not benchmarks. Tune them against your own won and lost deals.

ComponentSignalSuggested points
Fit (max 40)Industry and company size match your ICP+20
FitBuyer has a title that signed off on past deals+20
FitCompany too small or outside your market-20
Intent (max 50)Demo or pricing request+50
IntentPublic post asking for a tool like yours+30
IntentPricing page visit or comparison content viewed+15
IntentNewsletter signup or single blog visit+3
Freshness (max 10)Last meaningful action within 7 days+10
FreshnessNo meaningful action in 90 days-15
DisqualifiersCompetitor, student, job seeker or unsubscribed-50

Freshness is the part that often gets left out. Two leads with identical history are not equal if one acted yesterday and the other acted in spring.

What score should trigger a sales handoff?

Set the handoff threshold so that no single low-value action can reach it, but a direct buying action can. Then label leads by fit and intent separately so sales sees more than a single number.

Adobe’s guidance suggests starting with a threshold of 100 if you are new to scoring, requiring multiple interactions so one email open is never enough, and giving high-intent actions like a demo request a value equal to the threshold so they go straight to sales.

HubSpot’s score thresholds take a useful extra step. They label each record from A (high fit) to C (low fit) and from 1 (high engagement) to 3 (low engagement), so a low-fit but very engaged contact shows up as C1. That two-part label maps cleanly to actions:

Engagement 1 (high)Engagement 2 (medium)Engagement 3 (low)
Fit A (high)Route to sales todayPersonal outreach from an SDRNurture, watch for triggers
Fit B (medium)Qualify fast by phone or replyNurture with targeted contentNurture
Fit C (low)Check for a referral or a smaller planAutomated nurture onlyNo action

The grid also gives sales a reason to trust the score. An A1 lead is easy to explain. A raw score of 87 is not.

Why do lead scores go stale, and how does score decay help?

Scores go stale because most models only add points. A lead who downloaded three guides last year can still outrank someone who asked for pricing this morning. Score decay fixes this by reducing points as activity ages.

HubSpot lets you decay scores automatically based on how long ago a scored event happened, choosing both the percentage and the time period. Its worked example applies a 50% decay every three months to each rule. Negative points for inactivity, like the -10 for 90 days without interaction in Adobe’s example, do a similar job in tools that lack built-in decay.

Fit data goes stale too. People change jobs and companies change size, so the fit half of the score needs fresh records underneath it.

Where do public intent signals fit in a lead scoring model?

Most scoring models only see what happens on your own website and emails. They miss the buyer who is asking peers for recommendations somewhere else before ever visiting you. Public posts that ask for a solution are some of the strongest intent signals available, and they belong near the top of your intent points.

Think of a LinkedIn post asking for tool recommendations, or a Reddit thread where someone describes the exact problem your product solves. These are the kind of sales trigger events that tell you when to reach out, not only who to reach. FindyLeads monitors Reddit and LinkedIn for buying signals that match your product, surfaces those conversations, and drafts a reply you can edit before posting. Feeding those signals into your engagement score means a lead can earn intent points before they ever fill in a form.

Worked example: scoring three leads

This is an illustrative scenario, not a real customer. Using the Fit-Intent-Freshness template above, with a handoff threshold of 70:

LeadFitIntentFreshnessTotalAction
Ops lead at a target-size SaaS company who posted asking for tool recommendations yesterday+40+30+1080Route to sales today
Founder of a 4-person agency who requested a demo0 (+20 buyer title, -20 too small)+50+1060Qualify fast, check fit for a smaller plan
Marketing manager at a target account with three blog visits last spring+40+9-1534Nurture, watch for new triggers

Notice that the third lead looks good on fit alone. Freshness and real intent pull it down to where it belongs.

Frequently asked questions

What is a good lead scoring model for a small B2B team?

Start with a simple rules-based model that scores fit and intent separately, with fewer than fifteen rules in total. Build it from your won and lost deals, add negative points for poor fit and inactivity, and review it every quarter.

How is lead scoring different from lead qualification?

Lead qualification defines what a good lead looks like, often using a framework like BANT or MEDDIC. Lead scoring applies those criteria automatically across every lead so your team knows who to contact first.

What is score decay in lead scoring?

Score decay automatically reduces a lead's points as the activity that earned them gets older. It stops leads with old activity from outranking leads who are showing interest right now.

Should a demo request automatically qualify a lead?

In most models, yes. Adobe's Marketo guidance suggests giving high-intent actions like a demo request a point value equal to the handoff threshold so they go straight to sales, while the fit score tells the rep how to handle the call.

How often should you update a lead scoring model?

Review it at least once a quarter. Compare scores against deals that actually closed, ask sales which high-scoring leads were a waste of time, and adjust the weights and threshold.

What should you do this week?

Pull your last 20 closed-won and 20 closed-lost deals, list the fit traits and actions that separate them, and turn that into a two-part score with a decay rule and a review date. Then add one intent source your current model cannot see, such as public posts where buyers ask for a tool like yours. A small model that sales trusts beats a complex one they ignore.

Sources

Try it yourself

Find the people this playbook is about

Search 1 billion people and 450 million companies, then reach them while they are still looking. 7-day free trial, $0 due today.