Altra FCU

Efficiency and automation are the Altra-mate goals

Working with Zest AI, Altra built a new AI-based underwriting model, trained and validated on 50 months of prior loans. The team parsed thousands of raw bureau variables to select nearly 150 that predict delinquencies with far higher accuracy in a fully automated decisioning process. The results were impressive, increasing their automated approvals by 2.5x, and their automated decisioning by 1.3x.

The zest AI effect

6.7

increase in auto approvals for personal loans*

2.5

increase in auto approvals across all consumer portfolios

1.8

increase in auto decisioning for personal loans*

1.3

increase in auto decisioning across all consumer portfolios

At a glance

Problem

Altra Federal Credit Union's manual underwriting process and reliance on national scores were limiting their ability to scale, holding back their goal of doubling loan volume within 10 years while also making it harder to serve members in low-income communities.

SOLUTION

Altra partnered with Zest AI to build a new AI underwriting model trained on 50 months of prior loan data, selecting nearly 150 predictive features from thousands of bureau variables to power a fully automated, more accurate and inclusive decisioning process.

RESULTS

Automated approvals increased 2.5x overall and automated decisioning grew 1.3x, with personal loans seeing even stronger gains of 6.7x in automated approvals and 1.8x in automated decisioning.

Scaling up, the digital way

Altra Federal Credit Union was in the process of building a new digital lending platform and was looking to automate more decisions, with a goal of doubling loan volume within 10 years and reducing the amount of human interaction from application to approval. However, Altra’s manual underwriting process and reliance on national scores limited their team’s ability to grow, and serve more members in low income communities. They wanted to provide a better member experience, be more efficient, and say yes more, without taking on additional risk.

Getting it done, gaining efficiencies

Working with Zest, Altra built a new AI underwriting model, trained and validated on 50 months of prior loans. The team parsed thousands of raw bureau variables to select nearly 150 features that predict delinquencies with far higher accuracy in a fully automated decisioning process that is also more accurate and inclusive. Overall, Altra has increased its automated approvals by 2.5x, and their automated decisioning by 1.3x. Their personal loans portfolio particularly stands out, increasing automated approvals by 6.7x and automated decisioning by 1.8x.