How to set up lead scoring in a CRM
Lead scoring is only as good as its inputs and its calibration. A model built on gut weights and thin data misranks leads and erodes trust.
The disciplined build defines fit and engagement, feeds them complete data, and tunes the weights against what actually converted.
Short answer
Set up lead scoring by defining your ICP for fit, choosing the engagement signals that predict intent, scoring fit and engagement separately, enriching leads so fit can be computed, setting thresholds for action, and refining against actual conversions. Enrich for fit, watch real engagement, keep the two scores separate, and tune the model with outcomes.
Step by step
Define fit from your ICP
List the firmographic traits that mark a good-fit account and weight them, so fit reflects your real ideal customer.
Choose engagement signals
Pick the behaviors that predict intent - demo requests, pricing views, repeat visits, replies - and weight them by strength.
Score fit and engagement separately
Keep the two as distinct scores so you can tell a quiet good-fit lead from an active poor-fit one.
Enrich so fit can be computed
Fill firmographic data through enrichment, since fit cannot be scored on missing company attributes.
Set thresholds and refine on outcomes
Define score levels that trigger action, then tune the weights against which leads actually converted.
Tune against real conversions
A scoring model built on assumptions misranks leads until it is calibrated against reality. Periodically check which scores actually converted and adjust the weights, so the model reflects what predicts a real customer rather than what you guessed would. Calibration is what earns rep trust.
How Ardovo helps
Ardovo scores fit and engagement separately on enriched data and real activity, and Rook tunes the model against actual conversions. The score reflects what truly predicts a customer, so reps trust the ranking and work the right leads first.
Frequently asked questions
What data do you need for lead scoring?
Firmographic data for the fit score (industry, size, revenue, role) and behavioral data for engagement (visits, opens, replies, demo requests). Fit requires enriched company data, which is why enrichment is a prerequisite - you cannot score fit on missing firmographics.
How do you know if a lead scoring model is good?
Check whether high-scoring leads actually convert more than low-scoring ones. If they do not, the weights are miscalibrated. Tuning the model against real conversion outcomes - not assumptions - is what makes the score predictive and worth reps trusting.
Should lead scoring be one number or two?
Two is better: a fit score and an engagement score. One blended number hides whether a lead is a quiet good fit that needs nurturing or an active poor fit that is a bad use of time. Separate scores make the right action clear.