A new way of credit scoring using natural language processing and alternative data
Challenger credit bureau
We enabled more accurate credit decision making by lenders and broadened financial inclusion for consumers.
Increase in credit acceptance
Reduction in default rate
Background
A challenger credit bureau wanted to see how combining natural language processing with transactional data could enable more accurate and fairer credit scores, with personalised explanations to applicants automatically generated by AI.
Solution
We built a solution that generates accurate credit scores by predicting whether applicants will repay or default on a loan. The solution was trained on transactional banking data. Applicants also get personalised, detailed feedback so they can improve their chances of getting loans in the future.
Impact
This technology demonstrated to the credit bureau how they could boost loan approvals by 40% and reduce defaults by 46% through data-driven, accurate decision-making. The bureau can also make credit more accessible and decisions more transparent to consumers.
“Financial inclusion is a moral imperative and a commercial opportunity. We're proud to have helped our client achieve both.”