Arthur J. Gallagher & Co. is a global insurance brokerage and risk management firm where brokers are responsible for recommending insurance carriers that best match a client’s coverage needs and risk profile. This process is currently manual and relies heavily on individual broker experience, leading to inconsistent recommendations despite the availability of extensive historical data. To address this challenge, this project uses a two-stage machine learning framework that combines unsupervised clustering to identify carrier appetite profiles with a LightGBM recommendation model to rank insurance carriers based on a specific client’s needs. This framework provides brokers with more consistent, data-driven recommendations. The implementation of this solution could increase Gallagher’s revenue up to $2 million per year that are currently unrealized due to unsuccessful quoting outcomes.
Watch the team present this project at 00:08:06 in the session recording here.
Keywords: recommendation systems, insurance, insurance brokerage, clustering algorithms, LightGBM