Decisioning at the Edge: Policy Matching at Scale

A challenge at a global insurance company involves optimizing the assignment of online insurance policies to independent agencies. The previous manual, round-robin approach was inefficient, leading to the development of a new system using a lightweight integer programming model. Implemented with the PuLP Python library, the solution aims to maximize a productivity score based on agency performance metrics. The model operates in both batch and online modes, considering constraints like agency capacity, geographic eligibility, and fairness. This data-driven method provides a deterministic and auditable process for real-time policy assignment, moving away from subjective human judgment.