Case Study: Snap Finance achieves real-time lease approval and a 20% reduction in early-stage defaults with Domino Data Lab

Using predictive models to approve merchandise lease applications in real time

Snap Finance is a digital lender that offers merchandise lease-purchase agreements to consumers with poor credit, and it uses predictive models to approve leases on its website and in stores. As Snap moved to more sophisticated machine-learning models built in R and iterated on them rapidly, it needed a way to integrate those R models into its Java-based web app without constant engineering involvement, while meeting strict requirements for low latency, security, and reliability.

Using Domino’s API Endpoints, Snap publishes R models as one-click web services that its Java systems call via simple HTTP requests; Domino also handles secure, low-latency hosting and smooth version cutovers. Snap went live in three weeks, has deployed dozens of updates without engineering or downtime, and expects about a 20% reduction in early-stage defaults, improving both top- and bottom-line results.

Snap Finance

Tyler Hunt

Data Scientist