FINANCIAL SERVICE PROVIDERS heavily rely on revenue from loans, such as home, auto, or credit card loans. However, approving loans for the wrong borrowers can be costly. According to the Consumer Financial Protection Bureau, charge-offs in Q3 2024 reached 4.37%, with an additional 3.23% in delinquencies, representing billions in USD losses. To drive profitable loans and accelerate the adoption of their products, these providers often collaborate with third-party partners and pay commissions for new loans.
High loan acceptance can mean lost revenue
Third-party partners seek a high acceptance rate for their applicants to increase their revenue. To achieve this, the financial services provider would have to approve unqualified individuals, leading to revenue losses. A 1% change in the approval rate can impact monthly revenue by millions of dollars. The financial institution’s current methods group applicants by risk, which are easy to handle by business users but result in the rejection of qualified candidates and loss of revenue.
Factored integrated with the team
Factored partnered with the financial institution backed by our Centers of Excellence in Data Engineering, Data Analytics, Machine Learning and Software Engineering.
ORIGINATION MODEL
A set of parameterized rules to grant or deny a loan by evaluating different customer characteristics and generate a probability of future payment behavior.
COMPLIANCE APPROVAL RATE (CAR)
The proportion of approved customers related to all the applicants for the specific loan.
GROSS APPROVAL RATE (GAR)
Accounts for all the interactions with the customer during their application.
ORIGINATION MODEL
The result of the Data Science models made to anticipate the profitability of each customer based on their characteristics such as credit Bureau’s scores and credit report features.
COST OF ACQUISITION (CAC)
Marketing expenses associated with how much it costs to get an applicant to the loan.
RISK GROUPS
The Risk Groups were determined using a range of scores generated by the origination model.
Allowing us to assess the risk of each individual
We mapped these dynamic components to individual applications rather than the risk group to return a recommendation to approve or deny. Then we visualized the model trends showcasing different scenarios and the impact on acceptance rate, anticipated volume from the partners and increased losses from unqualified and approved loans.
Factored iterated multiple counterfactual scenarios in simulations to maximize the volume of approved loans by the financial institution.
Driving millions of USD in net profit
Our work improved the acceptance rate by a fraction of a percent, preventing the loss of thousands of qualified applicants from the riskiest target groups, leading to a reduction of millions of USD in Net Adjusted Charge-Offs annually and increasing net profit.
