Case Study: Leading Home Insurance Company achieves privacy-preserving, accurate home pricing with Mostly AI

Improving Home Insurance Pricing With Synthetic Geolocation Data

Leading Home Insurance Company needed to price policies across diverse U.S. climates but was blocked from using customer addresses by CCPA and HIPAA, preventing accurate risk assessment. They partnered with Mostly AI to provide synthetic geolocation data—synthetic home addresses tailored for modeling—so teams could enrich records without exposing personal data.

Mostly AI implemented a synthetization framework and delivered 15M synthetic home addresses that allowed lookups of five climate risk features from public databases; the pricing model trained on this synthetic data performed as well as the model trained on real data. The solution eliminated re‑identification risk, preserved 100% utility and original statistical dispersion, and shortened time‑to‑data from 6 months to 3 days (60x faster), unlocking new insights for pricing.