Lenders are increasingly integrating alternative data sources to improve underwriting accuracy and expand credit access. Personify Financial reported that incorporating verified bank transaction data increased approvals for underserved segments by up to 8% without raising credit costs. Meanwhile, Affirm’s transformer model, which analyzes detailed credit history patterns, generated 3.4% more completed purchases at comparable risk levels.
Plaid's LendScore model demonstrates the potential of these datasets, showing a 73% approval rate compared to 65% for traditional benchmarks at equivalent risk. The model also achieved 41% lower risk at the same approval rate and provided up to 5x more approvals for thin-file consumers. These tools address a significant market gap, as the Consumer Financial Protection Bureau estimates 25 million U.S. adults lack sufficient credit history.
Data quality remains a primary challenge for business lending. A survey of 350 financial executives revealed that 57% view inaccurate or incomplete records as a major underwriting obstacle. Consequently, microbusiness applications face a 77% approval rate, significantly lower than the 95% seen for large enterprises. About 19% of microbusiness applications are rejected due to insufficient information to verify creditworthiness.
To mitigate these issues, companies like Sage and Satago are leveraging structured accounting data for invoice finance underwriting. Their partnership provides lenders with real-time access to invoices, ledger data, and cash-flow information. This approach aims to provide a more holistic view of applicants, though the Federal Reserve notes that alternative models still face challenges regarding data quality, costs, and lack of testing across full business cycles.
