Field notes: the submission that looked fine until we pulled the roof

The application read clean. Four-unit frame building, built 2003, "composition shingle," two stories, no prior losses. The applicant's broker had filled in the fields the way brokers fill in fields: fast, confident, and from memory of what the building probably looks like rather than what it actually looks like today.

Nothing about it should have slowed anyone down. Construction type matched the occupancy. Year built lined up with the neighborhood. The loss run was quiet. On a busy Thursday, this is the file that goes through on the strength of the narrative alone, because the narrative is plausible and plausible is usually enough to keep the queue moving.

It's the roof that didn't match the story.

Pulled up against current aerial imagery, the building had a mansard addition over what used to be a flat rear section, built out sometime after the last inspection anyone had on file. Two slopes read as standing-seam metal, not shingle. And where the application said two stories, the roofline showed a third set in behind a parapet, the kind of setback that doesn't show from the street and doesn't show on a form unless somebody specifically asks the right question in the right way.

None of this is a dramatic reveal. It's the ordinary gap between what a building was when someone last looked at it and what it is now. Roofs get patched with whatever material was on the truck that week. Additions go up without anyone updating the file that matters to an insurer. A multifamily owner who self-manages six properties is not thinking about your construction-type field when the roofer quotes a job; they're thinking about the leak.

The pattern worth noting here is simpler than fraud. The form was never built to catch an honest gap, and the broker didn't lie so much as fill it out in good faith, based on what they remembered or what the last renewal said, while the building moved on without them.

What made this one worth flagging in the file wasn't the metal roof or the extra story on its own. It was that both showed up on the same structure, which usually means the same renovation event, which usually means somebody changed the risk without changing the paperwork. A standing-seam retrofit alone might just mean a better roof than the file gives credit for. A third story that reads as unreported occupancy is its own conversation. Together, on the same footprint, they're a reason to open the file before it closes, not after a claim does it for you.

This is the kind of thing that used to require someone physically driving the block, or at minimum a desk review generous enough in time to go pull county permit records and cross-reference them against an application that was never going to volunteer the discrepancy. Most shops don't have that kind of time per file, and the files where it would matter most don't announce themselves. The four-unit with the quiet loss run and the clean narrative is exactly the one nobody double-checks.

That's the gap a per-address roof read closes before bind, not after. Property Characteristics pulls roof geometry, material, and story count straight off current VHR imagery and hands back a structured record for the address, so the underwriting file has something to check the application against before the policy is written, not just a form that reads clean because nobody looked past it yet.

A clean narrative and a current roof are two different claims, and only one of them is verifiable before you bind. Worth checking the one you can.

If your desk sees enough of these to make a habit of it, it's worth seeing what Property Characteristics would have flagged on this file before it ever reached underwriting.