You built a lead scoring model. You set points for company size, engagement, budget authority. You tuned the thresholds. You rolled it out. And then your sales reps looked at it once and went back to their gut. This is not a training problem. It's not a communication problem. It's a signal problem. Your score is telling reps to pursue leads that they know won't close. The model doesn't match their intuition because the model was built on marketing theory, not on the deals that actually closed. The fix: rebuild your score from the deals reps already care about. Audit the last 50–100 closed deals. Find the patterns reps actually acted on . Assign points to those signals. Test with 2 reps. Iterate. Roll out. This post walks you through the exact steps and gives you a template rubric you can use today. Why reps kill your lead score Lead scoring is one of the most ignored tools in modern CRM. Studies show adoption rates between 15% and 35%, depending on the survey. But the problem isn't that reps are lazy or resistant to structure. The problem is that most scoring systems are built backwards. A typical model looks like this: Company fit: +10 points if revenue > $10M, +5 if in target industry, +3 if in target geography Buying signals: +20 for website demo signup, +15 for pricing page visit, +10 for content download Engagement: +5 per email open, +10 per email click, +3 per page view Authority: +15 if LinkedIn title contains 'VP', 'Director', or 'Head' This looks rigorous. It's not. Here's what happens: A lead gets a score of 65 because they hit the revenue threshold, downloaded a whitepaper, opened two emails, and have 'Manager' in their title. The system flags them as sales-ready. Your rep looks at the lead, sees the company is in retail, remembers the last three retail deals they lost in the first call, and marks the lead as 'not a fit.' Who's right? Your rep. The scoring model built on generic signals is useless when it collides with specific domain knowledge . Reps don't ignore the score because they're stubborn. They ignore it because it's wrong for their business. The audit: find the patterns reps actually traded on Start by pulling 50–100 of your most recent closed deals. For each one, ask your reps one question: "What signal made you take this deal seriously?" Not the final reason they won. The early signal that told them it was worth their time to engage. The thing that made them return the first outreach email. The thing that made them say 'yes' to a discovery call. You'll hear patterns like: "They came inbound from a specific partner" (not just any inbound— from that partner ) "They already use our competitor, so they know the category" "A peer company in their vertical adopted us last year" "They mentioned budget in the first email" "They asked about a specific feature we own" "The deal came from an existing customer's referral" These are intent signals —not 'they engaged with content,' but 'they made a choice that tells us they have a real problem to solve.' Document 20–30 of these patterns across your 10 reps (if you have 10). Look for the ones that appear in 60%+ of closed deals. These are your scoring anchors. Build a scoring system reps will actually use Instead of generic engagement points, you're going to score on the patterns you just found. Here's a real example from a SaaS company with 10 reps: Referral from customer: +40 points (closed deal rate 68%) Competitor already installed: +35 points (closed deal rate 52%) Budget mentioned in first response: +30 points (closed deal rate 45%) Specific feature request match: +25 points (closed deal rate 38%) Peer company in same vertical adopted in past 12 months: +20 points (closed deal rate 28%) Inbound from target partner: +25 points (closed deal rate 41%) Email open (first two only): +3 points each (closed deal rate 12%) Company revenue in target range: +5 points (closed deal rate 8%) Notice what changed: the high-leverage signals are worth much more. Generic engagement is nearly worthless. A lead with a referral from an existing customer gets 40 points—not because the model is generous, but because your data shows that signal correlates with 68% close rates. Set your threshold at 50 points. A lead needs to hit 50 points before it goes into the 'ready for sales' queue. Under 50, it stays in nurture or moves to a junior SDR. Key insight: Your scoring system should feel conservative to reps. If it flags 80% of leads as 'ready to sell,' reps will ignore it because they know better. If it flags 20%, reps will start to trust it. Test with 2 reps, iterate, then roll out Don't deploy to your entire team. Deploy to your two best reps—the ones who close the most, and the ones who are open to feedback. Give them the new scoring system for 2 weeks. Tell them: "We want to know which leads feel right and which feel wrong." Track: How many leads they touch (should be 60–70% of the old volume if the score is working) How many they move to discovery (should be