Property risk data is information about a building and its surroundings that an insurer uses to decide whether to insure the building and what to charge. It includes records of past losses, flood and storm maps, the age of the roof, and details of how the building was built.
An insurer cannot visit every home or factory it insures, so it buys this data. Some data companies pool the past claims of many insurers. Each insurer can then see how often a type of building in an area has suffered a loss. Other companies measure roofs from aerial imagery, or map flooded areas from satellite images. Reading location data with AI models in this way is called geospatial AI.
Take, for example, an insurer that is asked to cover a factory near Chennai for ₹40 crore. The underwriter, the employee who decides on the application, enters the address. The data comes back within seconds. The factory is 600 metres from a river that flooded in 2015. The building is 22 years old and has a metal roof. Similar factories in the area have made a claim about once every eight years. The underwriter sees all of this on the underwriter workbench and raises the flood part of the premium. The policy is then issued in the insurance core system.
The same data helps after a disaster. When a cyclone is forecast, the insurer can list every building it covers in the cyclone's path. A catastrophe model goes further. It estimates what one large storm or earthquake would cost the insurer across everything it covers.
Property risk data is one input to underwriting. Banks use similar data. For instance, a mortgage servicer checks whether a home with a loan on it lies in a flood zone. The largest providers mostly serve American insurers, and several of them build their products in Hyderabad and Delhi.