Vintage year vs. year built: what a footprint layer can tell you

If you've ever had to reconcile a "vintage year" field against a county assessor's "year built" record and gotten two different answers for the same parcel, you already know these terms aren't interchangeable. They get used loosely in submissions, and that looseness causes real problems when you're trying to assign a vulnerability curve.
Vintage year is a modeling convention, not a fact
Year built is, in theory, a discrete, recorded event: the date a certificate of occupancy was issued, or the date on the permit file. Vintage year, as most cat models use it, is a bucketed proxy meant to capture code-era construction practice. A structure coded "vintage 1995" is a statement about how it should behave under the codes in force around that time, a stand-in for construction practice rather than a record of when the foundation went in.
The gap between those two things is where a lot of submission data breaks down in ways that matter to curve selection. A schedule entry that says "year built: 1987" might be the date of original construction, the date of a prior appraisal, a renovation date somebody typed into the wrong column, or a default value a broker's system fills in when the field is blank. You can't always tell which from the number alone.
What a satellite footprint layer actually gives you
A footprint layer built from annual high-res imagery tells you what's physically standing, right now, across a territory. It gives you a building's shape, its rough size, and whether it's there at all compared to last year's pass or the year before that. Run the same coordinates against two or three years of imagery and you can see a parcel go from bare lot to slab to finished structure. That's a real construction event you can date to a window, sometimes a tight one if the cadence is annual and the build was fast.
What it doesn't give you is a permit date, a code-adoption year, or a material spec. Imagery shows you a roof and a footprint. It doesn't show you rebar schedules, anchorage details, or which edition of the building code the contractor pulled the permit under. Two structures can look identical from above and be a decade apart in actual construction practice, because the exterior form of a tilt-up warehouse or a standard tract home doesn't change much even as the codes underneath it do. A footprint layer is evidence of physical presence and physical change. It was never meant to stand in for the permit record.
Where this helps a renewal
The practical use case here is catching the places where the schedule is wrong in a way that matters, using the imagery as a check rather than a replacement for year built. If a structure shows up in this year's pass but not in imagery from three years back, that's a flag worth checking against the submission's stated vintage, because something was built or substantially rebuilt in a window the SOV might not reflect. If a parcel the schedule lists as "vacant" or "land only" clearly has a building on it now, that's a coverage gap, not a vintage dispute. And if a whole block of a territory shows heavy new construction that the underlying schedules haven't caught up to, that's a market-level signal about where exposure is accumulating faster than the self-reported data admits.
None of that tells you the exact year-built figure for a given parcel. It tells you where the built environment and the paper schedule have drifted apart, which is usually the more useful question going into a treaty renewal anyway. Exposure Mapping builds that current, consistent footprint layer from annual satellite passes, so you're checking your submission against what's standing on the ground rather than against whatever year someone last typed into a spreadsheet.
If your renewal exposure is built on schedules that haven't been checked against a current footprint layer in a while, that's worth fixing before the next submission cycle.