Building a Lead Scoring Model Without a Data Team

🟡 Series Difficulty: MEDIUM (Part 6 of 10) — Uses research skills from Part 2 and connects to MarketBetter's signal data. The most analytical post so far.

Every SDR knows the frustration: you've got 200 leads in your queue, and they all look the same. Same priority level. Same generic tags. No clear signal about who to call first.

So you do what every SDR does — you start at the top of the list and work your way down. Or you sort alphabetically. Or you go with gut instinct. None of these are strategies. They're survival mechanisms.

Meanwhile, the enterprise sales teams down the hall have sophisticated lead scoring models built by data teams, powered by Marketo or HubSpot, with algorithms that predict which leads are most likely to convert. You don't have that. You don't have a data team. You don't have a marketing ops person who can build predictive models. You have a CRM, a list of leads, and a quota.

Here's the good news: you can build a lead scoring model in 30 minutes using Claude Code. It won't be as sophisticated as a machine-learning-powered enterprise system. But it'll be 10x better than alphabetical sorting. And when you pair it with MarketBetter's daily playbook, you'll have a complete system for knowing exactly who to call first, every morning.

This is Part 6 of our Claude Code + MarketBetter series — the last of the Medium-level posts. In the Basic posts (Parts 1-3), you learned to research and write. In Parts 4 and 5, you built multi-step workflows for LinkedIn and competitive intel. Now you're going to do something more analytical: use Claude Code to build a system that makes decisions for you. You'll define scoring rules, apply them to data, and create a repeatable process that gets smarter over time.

If that sounds complex, don't worry. The Claude Code prompts are just as straightforward as the ones you've been using. You're just asking slightly more structured questions.

Let's build your scoring model.

What Is Lead Scoring (and Why Do You Need It)?

Lead scoring assigns a numerical value to each lead based on how likely they are to buy. Higher score = more likely to convert = call them first.

Simple concept. But most scoring models fail because they're either:

• Too complex — Built by data teams with 47 variables that nobody understands
• Too simple — "Enterprise = high priority" doesn't tell you anything useful
• Too static — Set once and never updated, even as your market changes
• Disconnected from action — Great model, but nobody uses it in their daily workflow

The model we're going to build avoids all of these traps. It uses three categories of signals, is easy to understand, and plugs directly into your MarketBetter daily playbook.

For a deeper dive on scoring best practices, check out our lead scoring best practices guide.

The Three Pillars of SDR Lead Scoring

Your scoring model is built on three pillars:

Pillar 1: Firmographic Fit (Does this company match our ICP?)

This is the "who are they?" question. It includes:

• Company size (employee count or revenue)
• Industry
• Geography
• Technology used
• Funding stage

Pillar 2: Behavioral Signals (Are they actively interested?)

This is the "what are they doing?" question:

• Website visits (especially high-intent pages like pricing)
• Email engagement (opens, clicks, replies)
• Content downloads
• Social media interactions
• Event attendance

Pillar 3: Timing Signals (Is now the right moment?)

This is the "when is the right time?" question:

• Recent funding rounds
• Leadership changes
• Job postings in relevant departments
• Competitor contract renewals
• Seasonal buying patterns

Each pillar contributes to a total score. The leads with the highest combined score get your attention first.

Step-by-Step: Building Your Model with Claude Code

Step 1: Define Your Ideal Customer Profile

Before you can score leads, you need to know what a great lead looks like. Ask Claude Code:

"Help me define my Ideal Customer Profile (ICP). I sell [your product] to [your market]. My best customers tend to be:

• Company size: [range]
• Industry: [industries]
• Typical buyer title: [titles]
• Common pain points: [pains]

Based on this, create a firmographic scoring rubric with a 0-30 point scale. Give me the exact criteria for each score level."

Claude Code returns something like:

Firmographic Scoring (0-30 points)

Criteria	Points	Details
Company Size	0-10	1-49 employees: 2pts, 50-200: 7pts, 201-500: 10pts, 500-1000: 8pts, 1000+: 5pts
Industry	0-10	SaaS/Tech: 10pts, Financial Services: 8pts, Healthcare: 6pts, Manufacturing: 3pts, Other: 1pt
Geography	0-5	US: 5pts, UK/Canada: 4pts, Western EU: 3pts, Other: 1pt
Funding Stage	0-5	Series A-C: 5pts, Seed: 3pts, Bootstrapped: 2pts, Public: 2pts

Notice how the scoring reflects YOUR specific ICP. A 200-person SaaS company in the US scores higher than a 5,000-person manufacturer in Asia — because that's who buys from you.

Step 2: Build the Behavioral Scoring Component

Now add the engagement signals. This is where MarketBetter's data becomes critical:

"Now create a behavioral scoring rubric (0-40 points) based on these engagement signals I can track:

• Website visits (from MarketBetter visitor identification)
• Pages visited (pricing page, case studies, product pages)
• Visit frequency (one-time vs. return visitor)
• Email engagement (opens, clicks, replies)
• LinkedIn engagement (profile views, connection accepts, post interactions)

Weight the signals by purchase intent. A pricing page visit is more valuable than a blog page visit."

Claude Code returns:

Behavioral Scoring (0-40 points)

Signal	Points	Details
Pricing page visit	10	Single strongest buying signal
Case study/testimonial page	7	Evaluating social proof
Product/feature pages	5	Active research phase
Blog/content visit	2	Awareness stage
Return visitor (2+ sessions)	8	Sustained interest
Multi-page session (3+ pages)	5	Deep engagement
Email opened (2+ times)	3	Interest but not action
Email link clicked	5	Active engagement
Email replied	8	Direct interest
LinkedIn connection accepted	3	Openness to conversation

Step 3: Build the Timing Scoring Component

Finally, add signals that indicate the timing is right:

"Create a timing/trigger scoring rubric (0-30 points) based on these signals:

• Recent funding announcement
• Executive leadership changes
• Job postings in relevant departments
• Company expansion/new office
• Technology changes or migrations
• Contract renewal season (if known)

Weight by urgency of the buying window."

Claude Code returns:

Timing Scoring (0-30 points)

Signal	Points	Details
New funding (last 60 days)	8	Budget available, growth mandate
New CRO/VP Sales (last 90 days)	7	New leaders bring new tools
Hiring SDRs/AEs (active postings)	6	Scaling sales = needs tools
Hiring demand gen/marketing	5	Building pipeline infrastructure
Technology migration announced	6	Open to new vendors
Competitor contract likely up for renewal	5	Evaluation window
Expansion/new market entry	4	Growing pains = new needs

Step 4: Score Your Existing Leads

Now apply the model. Export your lead list from your CRM and feed it to Claude Code:

"I have a list of 100 leads. Apply this scoring model to each one:

[paste your scoring rubrics]

For each lead, I have:

• Company name, size, industry, geography
• Website visit data from MarketBetter (pages visited, frequency)
• Email engagement data (opens, clicks, replies)
• Any known trigger events

Score each lead across all three pillars, calculate the total, and rank them from highest to lowest. Group them into tiers:

• Hot (70-100): Call immediately
• Warm (40-69): Prioritize this week
• Cool (20-39): Nurture sequence
• Cold (0-19): Low priority

Here's the data: [paste your lead list with available data]"

In 2-3 minutes, you have a fully scored, prioritized lead list. No data team required.

Using MarketBetter's Daily Playbook as the Execution Layer

A scoring model is useless if it doesn't change your daily behavior. Here's how to connect your Claude Code scoring model to your MarketBetter workflow:

The Morning Ritual (10 minutes)

1. Check MarketBetter's daily playbook — New website visitors, return visitors, engaged prospects
2. Apply your scoring model — New behavioral signals from overnight activity change scores
3. Identify your Hot tier — These are your first calls of the day
4. Identify new entrants to Warm tier — Prospects who were Cool but just visited your pricing page. They jumped tiers overnight.
5. Execute — Start with the highest-scored leads and work down

Signal-Triggered Score Updates

MarketBetter sends you real-time signals throughout the day. Each signal should update your mental scoring:

• Prospect visited pricing page → +10 points. If they were Warm, they're now Hot. Call them.
• Prospect opened your email 3 times → +5 points. They're interested. Send a follow-up.
• Prospect visited your site from a new device → +3 points. They might be sharing your site with colleagues. Multi-stakeholder interest.
• Cold lead returned to your site → Re-score them entirely. They might have jumped from Cold to Warm in one visit. (More on re-engagement in Part 9.)

Automated Scoring with MarketBetter

MarketBetter's built-in engagement tracking does much of the behavioral scoring automatically. Your Claude Code model handles the firmographic and timing scoring that MarketBetter doesn't cover. Together, they give you a complete picture.

For more on how intent data drives this process, read our guide to what intent data is and how it drives growth.

Refining Your Model Over Time

Your first scoring model won't be perfect. That's fine. Here's how to improve it:

Monthly Review (15 minutes)

"Here are my last month's results:

• 15 leads scored Hot → 8 converted to meetings (53%)
• 30 leads scored Warm → 6 converted to meetings (20%)
• 45 leads scored Cool → 2 converted to meetings (4%)
• 10 leads scored Cold → 0 converted to meetings (0%)

Also, 3 meetings came from leads scored Cool or Cold. Here's what those leads had in common: [details]

Based on this data, what adjustments should I make to my scoring model? Are any signals over- or under-weighted?"

Claude Code will analyze the conversion data and suggest specific adjustments. Maybe pricing page visits should be worth 15 points instead of 10. Maybe industry scoring needs recalibration. Make the adjustments and run the updated model.

The Feedback Loop

Over 3-6 months, your scoring model gets increasingly accurate because you're refining it based on actual conversion data. This is essentially what data teams do with machine learning — just simpler and driven by your domain expertise instead of algorithms.

Advanced: Multi-Persona Scoring

If you sell to multiple buyer personas, you might need different scoring models for each:

"I sell to two different personas:

Persona 1: VP of Sales (cares about pipeline and team productivity) Persona 2: RevOps Leader (cares about data quality and tech stack efficiency)

Create separate behavioral scoring rubrics for each persona. A VP of Sales visiting a case study page is different from a RevOps leader visiting an integration page — weight them differently."

This gives you nuanced prioritization. A RevOps leader on your integrations page might score higher than a VP of Sales on your blog — even though the VP is the more senior title — because the RevOps behavior signals active evaluation.

Common Scoring Mistakes to Avoid

1. Over-weighting title/seniority — A Director who's actively researching is more valuable than a VP who isn't
2. Ignoring negative signals — Unsubscribes, bounced emails, and "not interested" replies should decrease scores
3. Scoring once and forgetting — Scores should be dynamic, updated with every new signal
4. Too many tiers — Hot/Warm/Cool/Cold is enough. Don't create 10 tiers that nobody can remember
5. Ignoring the denominator — If your Hot leads aren't converting at a higher rate than Warm leads, your model isn't working

Free Tool

Try our AI Lead Generator — find verified LinkedIn leads for any company instantly. No signup required.

Try This Today

Here's your concrete action item:

1. Open Claude Code and use the prompts from Steps 1-3 above to build your scoring rubrics
2. Pick 20 leads from your current queue
3. Score them manually using your new model (estimate where you can)
4. Sort them by score and compare the order to how you would have prioritized them with gut instinct
5. Work the list in score order for one week and track your results

Most SDRs find that their intuition was right about 60-70% of the time. A scoring model gets you to 80-90%. That 20-30% improvement in prioritization translates directly to more meetings with less effort.

---

This is Part 6 (🟡 Medium) of our 10-part series. You've completed the Medium tier! Next up: Part 7: CRM Cleanup in Minutes → — your first Advanced-level post.

MarketBetter's daily playbook surfaces the behavioral signals that power your lead scores. Book a demo to see how it works.