Case Study: Major Global Insurance Company achieves 83% risk-model accuracy and 10x faster underwriting with Cognizant

A Flood of Data Informs Underwriting Using AI

A global reinsurer wanted a clearer, faster way to assess U.S. flood risk and reduce time-consuming manual underwriting as it evaluated opportunities to reinsure tranches and write individual flood policies. With publicly released National Flood Insurance Program maps, census and housing data, and the client’s claims history, the challenge was to model risk at granular geography (Zip/Zip+4) and expand the data sources informing pricing and acceptance decisions.

Cognizant built an AI-driven underwriting solution that combined GIS and NFIP data with the client’s claims history, using R, ArcGIS, Bayesian analysis and NLP to extract and synthesize geospatial and document data and present risks in an RShiny dashboard. The model delivered granular risk scores and a mapped market view with 83% accuracy, cut underwriting throughput time tenfold and improved case acceptance by 25%.