How to triage a renewal book by roof risk without inspecting every address
A renewal book with ten thousand homeowner policies doesn't get ten thousand roof inspections. Nobody has the field capacity for that, and the loss history doesn't justify it on every address anyway. What you actually need is a way to sort the book before you decide where an inspector, a drone, or a non-renewal letter is worth the spend.
That's triage, not inspection. The goal is separating the addresses that need a closer look from the ones that don't, so your limited field hours go where the risk sits.
What you're sorting on
Age is the number most underwriting guidelines already lean on, and it's a reasonable starting filter. But roof age alone misses a lot. A 22-year-old roof that was re-coated and well maintained can outperform a 9-year-old roof that took hail last season and got patched instead of replaced. Age tells you when the roof was installed. It doesn't tell you what's happened to it since.
The things that separate a low-risk roof from one headed for a claim are visible on the roof plane itself: staining that suggests trapped moisture or a failing seal, granule loss that shows the surface is wearing thin, patching that means something already failed once, and tarps, which are about as plain a signal as underwriting gets. None of these require walking the roof to see. They show up in a decent overhead image.
Building the sort without a truck roll
Most books already have the address list and the renewal dates. What's missing is a condition layer you can put next to it without sending someone out to every property. That's where very high resolution aerial or satellite imagery earns its place in the workflow, read at a fine enough ground sample distance (under 0.3 m) to tell staining from shadow and a patch from a vent.
Pull that imagery once a year, ahead of your renewal cycle, and score each address on the handful of conditions that matter to your guidelines. Sort the output against your own thresholds. The roofs that come back clean skip the queue entirely. The ones flagged for heavy staining, visible patching, or a tarp move to the top of your non-renewal review or field-inspection list. Everything in between gets a second look at renewal time instead of a blanket order.
This is the part that saves field hours: you're not inspecting the book, you're inspecting the fraction of the book the imagery already told you to worry about. A team running this well might cut a field order list from thousands of addresses down to the few hundred that show a real condition signal, which is the whole point of the exercise.
Where it still needs eyes on the roof
A desk-level roof read is a sorting tool, not a replacement for inspection on the addresses that need one. If a roof comes back flagged for a tarp or a clear patch job, that's a candidate for a field visit or a non-renewal conversation, not an automatic decline. Imagery tells you where to look. A licensed inspector or adjuster still makes the underwriting call on what they find once someone's there.
The same logic applies on the clean side. A roof with no visible staining, no patching, and no tarp isn't guaranteed loss-free. It's just not showing the kind of surface damage that correlates with one. Treat the sort as a confidence ranking across the book, not a pass/fail stamp on any single address.
Running it as a yearly habit, not a one-off project
The value compounds when you run the same read every year ahead of renewal: a roof that was clean last cycle and shows new staining or a fresh patch this cycle gets flagged for that change as well as its condition. That year-over-year comparison is often more useful to an underwriting manager than any single snapshot, because it catches deterioration before it shows up as a claim.
Roof Scoring builds exactly that annual, per-address roof read for homeowner renewal books, sorted so you can run it straight against your own guidelines. If your non-renewal review still starts with a spreadsheet and a prayer, it's worth seeing what a year of imagery-based sorting looks like on your book.