Challenge
Decisions across credit, collections and fraud ran on fixed rules, which made them slow, inconsistent and costly. The bank held the data it needed, but that data was not shaping decisions in real time.
// Work 10
Moving a top-three UAE bank from rules-based to model-led decisioning
// Introduction
One of the UAE's three largest banks needed faster and more consistent decisions across credit, collections and fraud. Holly and the team led the bank's shift to model-led decisioning using machine learning.
What we did
The team embedded machine learning into the bank's credit, collections, fraud and risk models. It built predictive analytics across the customer lifecycle, covering propensity to skip, next-best offer and hyper-personalisation. A new data lake gave the bank a 360-degree view of each customer and supported real-time decisioning.
Outcome
Decisions became faster and more consistent. Predictive interventions improved collections, fraud detection became stronger, and the bank reduced manual processing and cost.
*Some case studies describe work delivered before forming Beacon58.