AI lead scoring is one of the few production-grade AI applications in mortgage that consistently produces measurable ROI. Done right, it lifts closed-loan rate 15-25% on the same lead volume by changing which leads get worked first.
How the model works
The model trains on the mortgage company’s historical data: leads that closed, leads that didn’t, and the attributes of each. It identifies which lead attributes correlate with closed loans (geography, income range, loan amount, lead source, response speed, document upload behavior) and assigns weights to each. New leads get scored against the model and ranked by close probability.
Training requirements
- 200-500+ historical funded loans for training
- Equal or greater number of leads that didn’t close (negative class)
- 12+ months of behavioral data on those leads
- Clean attribution between lead source and closed-loan outcome
For brand-new operations or LOs without sufficient history, AI lead scoring is not feasible. The alternative is rule-based scoring (response time + lead source + stated loan amount).
Compliance constraints
Lead scoring models cannot use protected-class attributes (race, ethnicity, age, gender, marital status, religion, disability) or proxies for them. ECOA and HMDA compliance limit what features the model can include.
