Lead Scoring
Lead Scoring is A numerical model that ranks prospects by fit (firmographic attributes) and engagement (behavioral signals). RevOps builds and tunes the model, and the MQL threshold determines when leads are handed to sales.
Lead scoring is the bridge between marketing activity and sales attention. It takes raw engagement data and firmographic fit, assigns numerical values, and produces a score that answers one question: is this lead worth a rep's time right now? Get it right and sales focuses on high-probability prospects. Get it wrong and you either waste rep time on junk leads or let good ones slip through.
The model has two dimensions. Fit scoring evaluates who the lead is: company size, industry, title, tech stack, geography. These are the attributes that match your ICP. Engagement scoring evaluates what they've done: pages visited, content downloaded, emails clicked, events attended, product trials started. High fit plus high engagement equals a hot lead.
Building and Tuning the Model
- Start simple: 5-7 fit criteria, 5-7 engagement triggers. Complex models with 40 variables are impossible to debug and rarely outperform simple ones.
- Weight by conversion correlation: Analyze which attributes and behaviors actually predict closed deals, not just what seems intuitively important.
- Set score decay: A lead who visited your pricing page 6 months ago isn't the same as one who visited yesterday. Engagement scores should decay over time.
- Validate quarterly: Check if high-scoring leads convert to SQL at a meaningfully higher rate than low-scoring ones. If not, recalibrate.
The MQL threshold is the score at which a lead is deemed qualified for sales. RevOps owns this number and adjusts it based on sales capacity and conversion data. Too low and sales drowns in noise. Too high and viable prospects go cold waiting.
Frequently Asked Questions
What's the biggest mistake in lead scoring?
Building a complex model and never validating it. Most lead scoring models are set-and-forget. If you aren't regularly checking whether high-scoring leads actually convert at higher rates than low-scoring ones, your model is decoration, not a decision tool. The second biggest mistake is weighting content downloads too heavily. Downloading a whitepaper doesn't mean someone wants to buy.
Should lead scoring use AI or manual rules?
Start with manual rules. You need to understand the logic before you automate it. Once you have 6-12 months of conversion data, predictive scoring models (offered by most MAPs and tools like 6sense) can surface patterns humans miss. But even with AI scoring, RevOps needs to understand what the model weights heavily and validate outputs against reality.
Get Weekly RevOps Intelligence
Salary benchmarks, tool reviews, and job market insights for revenue operations leaders. Every week.