All work
182014NETSOL Technologies

Lease Application Lead Scoring

Scoring lease applications so sales chases the right ones — +18% closed-lease yield, decisions in 8 hours not 36.

Before

Sales teams at OEM finance clients were chasing low-probability leases, burning cycle time on poor-fit applicants.

After

A logistic-regression scoring model on application and bureau features, exposed as an API into the leasing workflow, improved closed-lease yield by 18%, cut average time-to-decision from 36 hours to 8, and was deployed at two international leasing customers.

Challenge

Sales teams at OEM finance clients were chasing low-probability leases, eating up cycle time on poor-fit applicants.

Approach

Logistic-regression scoring model on application + bureau features, exposed as an API into the leasing workflow.

How it was built

  1. 1

    Feature engineering

    Phase 1

    Combined application data with credit-bureau features into a signal set that actually predicts whether a lease closes.

  2. 2

    Interpretable model

    Phase 2

    Chose logistic regression so underwriters could see why an application scored the way it did — trust matters in finance.

  3. 3

    API into the workflow

    Phase 3

    Exposed scoring as an API embedded in the leasing workflow, so the score arrives where the decision is made — cutting time-to-decision to 8 hours.

Key architecture decisions

Interpretable model over a black box

Why · In lease underwriting, an explainable score is adoptable; an unexplained one isn't. Logistic regression traded a little accuracy for the trust that drove adoption.

Score inside the existing workflow

Why · A model no one sees changes nothing. Embedding it as an API in the leasing flow is what turned it into an 18% yield lift.

Impact

  • Improved closed-lease yield by 18%
  • Cut average time-to-decision from 36 hours to 8 hours
  • Deployed at two international leasing customers
+18%
yield
-77%
decision
2 OEMs
customers

What I'd tell someone building this

  • 01 · In regulated finance, interpretability often beats a marginal accuracy gain.
  • 02 · Deliver the prediction where the decision is made, not in a separate report.
  • 03 · Good features beat fancy models — most of the lift came from the bureau data.

Tech stack

R.NETSQL ServerREST API

Ask anything about Lease Application Lead Scoring

AI scoped to this project · GPT-OSS 120B