How to Build a Lead Scoring Model That Sales and Marketing Both Trust

The Root Cause of Sales-Marketing Misalignment

Marketing celebrates MQL volume. Sales complains the leads are garbage. Sound familiar? The problem usually isn't the leads themselves — it's a scoring model that doesn't reflect what actually predicts a closed deal.

Here's how to fix it.

Start With Closed-Won Data

Open your CRM and filter for closed-won deals from the last 12–18 months. For each deal, note:

• Company size and industry
• Lead source
• First action the contact took (demo request, content download, webinar, etc.)
• Number of website visits before converting
• Job title of the contact
• Time from first touch to close

This is your ground truth. Your scoring model should weight attributes that appear most frequently in closed-won deals.

Two Dimensions of Lead Scoring

Effective lead scoring has two axes:

Fit score (who they are)

• Company size matches ICP: +20
• Industry matches ICP: +15
• Job title is your buyer persona: +25
• Company uses key tech signals: +10
• Geography matches: +5

Engagement score (what they've done)

• Requested a demo: +40
• Visited pricing page: +30
• Opened 3+ emails: +15
• Downloaded a high-intent asset: +20
• Attended a webinar: +20
• Visited careers page: -10 (likely a job seeker)

Calibrating the Thresholds

Once you have your scoring dimensions, set thresholds that determine lead status:

• 0–39: Nurture — not ready for sales
• 40–69: Marketing Qualified Lead (MQL) — marketing continues engagement
• 70–89: Sales Qualified Lead (SQL) — sales outreach begins
• 90+: High Priority — immediate follow-up required

Test these thresholds against historical data. If most of your closed-won deals scored above 70, your thresholds are in the right place.

Getting Sales Buy-In

This is the part most teams skip — and why scoring models fail.

• Include sales in the design process. Have reps score 20 recent leads manually, then compare to the model. Where do they disagree?
• Show the data. Present the closed-won analysis. Let the numbers make the argument.
• Start with a pilot. Run the model alongside existing processes for 30 days before fully switching over.
• Create a feedback loop. Let reps flag leads that scored high but were clearly unqualified. Use that data to refine the model.

Automate and Maintain

Build the scoring logic directly into your CRM or MAP (HubSpot, Salesforce, Marketo). Set up alerts when leads cross key thresholds.

Then review the model quarterly. Look at whether high-scoring leads are actually converting at higher rates. Adjust weights based on what you learn.

A lead scoring model is never finished — it gets better the more you feed it real-world outcomes.