Reconciling structure counts: cedant submission vs satellite imagery

Straight-down aerial view of a suburban grid, houses and yards laid out along crossing streets.

Every renewal season produces the same conversation. The cedant's bordereau says a given treaty territory holds 4,200 structures. Your underwriters want to know if that number is still true. Nobody on the submission side has driven the territory in three years, and the schedule was probably rolled forward from last year's schedule, which was rolled forward from the year before that. At some point you need a second count that didn't come from the cedant, and you need to know where the two numbers disagree and why.

Here's roughly how that reconciliation goes when you're doing it properly, not just sanity-checking a total.

Start with totals, then go granular

The first pass is simple arithmetic: does the submitted count fall inside a plausible range for the footprint you're independently counting? If the cedant says 4,200 and an imagery-derived count for the same boundary comes back at 4,600, that 400-structure gap is worth a look before you touch a single policy record. New construction the schedule hasn't caught up with could explain it. So could outbuildings and ancillary structures the cedant never captured because nobody asked their agents to. Or it could be a boundary mismatch, where the treaty territory and the imagery footprint aren't actually describing the same polygon.

That last one trips up more reconciliations than anything else. Before you draw conclusions about exposure growth, confirm the cedant's CRESTA zones or ZIP aggregates line up with whatever boundary the structure count was run against. A count that looks 15% high because it includes a neighboring zone isn't telling you anything about the cedant's book.

Once totals are in the same ballpark, granularity is where the real reconciliation work happens. Break both counts down by sub-area, by construction class if the imagery pass captured rough form, by anything that lets you compare like against like rather than two blunt totals. A cedant that's flat at the territory level but up 20% in one coastal sub-area is a different story than uniform growth, and it's the kind of thing that changes how you think about accumulation in that corner of the treaty.

What a gap actually tells you

A mismatch between submission and imagery doesn't automatically mean the cedant lied. Submitted schedules go stale for ordinary reasons. Infill construction between survey cycles doesn't show up until someone re-walks the book. Self-reported counts from smaller agents can lag actual exposure by a renewal cycle or two, especially in markets that build fast. And sometimes the imagery count is wrong in its own way, picking up structures that are sheds or carports rather than anything a treaty should be rating.

What the gap does tell you is where to ask questions before you bind. If the independent count is consistently higher than the submission across several sub-areas, that's a pattern worth raising with the cedant directly, not something to quietly absorb into your loss pick. If it's consistently lower, you may be looking at a submission padded with structures that no longer exist, demolished stock that never got pulled off the schedule.

Either way, the point of running an independent structure count isn't to replace the cedant's submission. It's to give your submission QA process a number that didn't come from the same source as the thing you're checking. That's the whole value of it: an input that's current and built from imagery rather than carried forward from last year's file.

If you're building that second number from annual high-resolution imagery rather than stitching it together manually, Exposure Mapping is built for exactly this: a current structure footprint and count layer for a treaty territory, generated from satellite imagery rather than self-reported schedules, that drops straight into your reconciliation workflow instead of another spreadsheet to chase down.

Worth running before the next renewal, not after the gap shows up in a loss.

Get started

← Back to the blog