Billion‑dollar underwriting data gap exposed

Commercial property underwriting is turning into a visibility fight. The basic problem is simple — insurers are still pricing and selecting risks with stale submissions, incomplete statements of values, and inspection cycles that move slower than the buildings themselves. That gap has become expensive enough that vendors, carriers, and regulators are now all treating external data as core infrastructure, not a side tool. The story here is not one flashy AI launch. It’s the industry admitting that missing evidence is showing up directly in loss ratios. ### What is the “data gap” here? Underwriters often get a clean address, a broker submission, and maybe some historical loss data. But a property can change fast — roof condition worsens, occupancy shifts, maintenance slips, a pool appears, a mixed-use exposure creeps in. If those facts are missing at bind time, the policy can be priced like one building and behave like another. That is the gap carriers keep talking about. ### Why is property insurance feeling it so hard? Because property losses have become more localized and more volatile. A ZIP code view is not enough when one roof is intact and the one next door is already failing. In habitational business — apartments and condos — Verisk described a $22 billion premium segment where losses have climbed an average of 10% a year over five years, pushing some carriers to pull back or move business into excess and surplus lines. (carriermanagement.com) ### Why do roofs keep coming up? Because roofs are the easiest example of hidden condition risk that turns into very visible claims. CAPE Analytics says severe-condition roofs show 3.9x higher loss ratios than excellent-condition roofs. In one example, its imagery and models flagged severe deterioration, an underwriter declined the risk, and the roof later collapsed in a claim believed to be above $100,000. That is the whole thesis in miniature — missing condition data is not abstract. It changes bind decisions and claim outcomes. (carriermanagement.com) ### So what are carriers using instead? Mostly remote evidence layers — aerial imagery, geospatial signals, permit data, business occupancy data, and AI models that turn raw images into underwriting flags. The point is not to replace every inspection. It is to decide where an inspection is actually worth sending. Verisk is packaging permit records, aerial-detected recreational exposures, occupancy clues, and core building attributes into targeted reports. CAPE is pushing “condition-aware underwriting,” basically scoring what the property looks like now rather than trusting what the file said months ago. (capeanalytics.com) ### Is this already mainstream? Not really — and that is part of the opportunity. EarthDaily said that at an InsurTech NY panel in late March 2026, only about 30% of insurers were using geospatial and location intelligence in underwriting. So the market is in an awkward middle stage: the tools are good enough to matter, but adoption is still uneven enough that a lot of underwriting decisions are being made half-blind. (carriermanagement.com) ### What’s the catch? Better data does not just improve pricing. It can also shrink the set of risks a carrier is willing to write. EarthDaily’s point was blunt — as visibility improves, underwriting boundaries get tighter and more explicit. That is great for carrier discipline, but it can also mean more properties get declined, repriced, or pushed into harder markets. In other words, precision helps insurers first. It does not automatically help availability. (earthdaily.com) ### Why are regulators stepping in? Because if insurers are going to use aerial imagery and AI-driven signals for underwriting or nonrenewals, states want guardrails. Colorado, Massachusetts, and Michigan have all issued guidance saying insurers can use aerial imagery, but they need accuracy, explainability, and additional review when images are not conclusive. Cosmetic issues alone should not trigger adverse action, and ambiguous cases may require physical inspection or contractor review. (earthdaily.com) ### Bottom line? The underwriting story is shifting from “who has the best appetite” to “who can see the risk clearly enough to act before the loss.” AI matters, but the deeper change is more boring and more important — standardized external evidence is moving into the core underwriting workflow. Carriers that can connect those signals cleanly will price faster and walk away sooner. Carriers that cannot will keep learning about bad risks the expensive way. (doi.colorado.gov)