Insurers have always wanted to know more about the properties they cover than a questionnaire can tell them. Roof age, surrounding vegetation, proximity to hazards, and undisclosed modifications all shape risk — and all are visible from above, well before they show up in a claim.
The limits of declared information
Underwriting still leans heavily on what the applicant declares. That information is often outdated, sometimes incomplete, and occasionally wrong. Re-inspecting properties at scale is too expensive to do routinely, so portfolios drift away from their underwritten assumptions over time.
Structuring the signal
A photograph does not help an actuary. A score does. The process is to detect features that correlate with loss — roof condition, defensible space, structure count, water proximity — then weight them into a consistent, explainable metric. The same property assessed twice should yield the same score; two similar properties should score alike.
Defensibility matters as much as accuracy: a score you cannot explain is a score you cannot use in a regulated decision.
From underwriting to portfolio monitoring
The same pipeline serves two jobs. At underwriting, it enriches a single application with objective ground truth. Across a book, it monitors continuously — flagging the property where vegetation has grown into a wildfire risk, or where a roof has visibly degraded, so the insurer can act before the loss rather than after the claim.
None of this replaces the underwriter. It gives them an independent, repeatable evidence base — and a record they can stand behind if the decision is ever questioned.