Custom Decision Engine Platform Helping Fintech to Scale Up

Project Background

Amplifi Capital, 2024 Scale-up of the Year of Fintech Awards London winner, is one of the most successful near-prime loan providers in the United Kingdom.

Anticipating growth in the number of loan offers, they examined their end-to-end approval process. They identified that their current box-solution decision engine might not support future business needs.

To support this growth and gain an advantage over competitors in the United Kingdom’s market, the client has decided to be less dependent on the supplier’s capacities and roadmap. They chose to take charge of the pricing model and wanted to be more flexible when adopting new external and internal services into the decision process.

Challenge

The United Kingdom’s market for unsecured personal loans is very competitive. As the end customer typically receives tens of online loan offers in real time, the success of the lending financial institution depends on lightning-fast and flawless data evaluation.

The most difficult challenge for the project was ensuring a smooth transition from the legacy solution. As time was a critical factor, it was decided that other back-bone systems would remain unchanged, requiring the new solution to be fully compatible with the entire environment at the data level. The decision logic also underwent extensive testing to ensure it produced the same results as the legacy solution.

Business Needs

The solution needed to meet the following requirements:

• Handle significant growth in the number of loan quotes
• Reduce unit costs per processed quote
• Support faster time to market for decision logic changes
• Provide a flexible solution that allows the integration of advanced decision-making using Machine Learning

Solution & Results

The essential part of the solution is a flexible application enabling business analysts to create and adjust decision engine logic without involving IT specialists. By employing a no-code approach, business analysts can release new decision engine versions and make necessary changes to the decision logic quickly, efficiently and reactively.

The time to market for decision logic changes has been further reduced by introducing automated testing, using hundreds of test data sets to check the quality of each delivery.

Additionally, the implementation has enhanced product launch. The legacy box-solution decision engine, with more than 2500 decision steps, was successfully decommissioned. The use of Machine Learning models simplified the decision logic, leading to cost savings on external services and improved response times for loan aggregators, reducing the average response time from 5 seconds to 2 seconds.

With the new solution, Amplifi has grown from 1 million requests per month to more than 2 million requests per month without any performance issues.