What is exposure data quality grading, and why do cedants get scored A through D?

Suburban houses of varying styles lined up along the edge of an open field under a cloudy sky.

Every submission that lands on a modeller's desk gets sized up fast. Before anyone runs a single event set, someone is asking: how good is this COPE data, really? That question gets answered with a letter. A, B, C, or D. Cedants who've been through a few renewals know the grade shows up in the submission notes, and they know it affects more than how the actuary feels about the file.

What the grade actually scores

The A through D scale (some shops use numeric tiers instead, 1 through 5) is a shorthand for how complete and how verified the primary characteristics are: construction class, occupancy, year built, number of stories, square footage, and the geocoding precision behind each location. A clean A-grade schedule has street-level geocoding, confirmed construction and occupancy codes, and a year-built field that isn't full of 1900 placeholders. A D-grade schedule is the opposite: ZIP-centroid locations, occupancy inferred from a SIC code, and a construction field that's mostly "other" because nobody filled it in.

The grade measures how much of the schedule is actual reported fact versus how much the model has to fill in with defaults and secondary modifiers, not the cedant's honesty. A broker-compiled schedule of values that's three years old and was never updated after a renovation program will grade low even if every number on it was true the day it was submitted.

Why the letter moves the loss number

This is the part that gets underweighted on the cedant side. A downgrade from B to C doesn't just sit in a footnote. It triggers wider uncertainty bands, pushes the model toward conservative construction and occupancy defaults, and in a lot of shops it adds a load to the loss cost before the underwriter even sees the quote. On a large commercial or industrial treaty, that load can move the technical price by a meaningful margin. Brokers who've watched a renewal price come in higher than expected, with nothing in the loss history to explain it, are often looking at a data quality penalty they didn't know was being applied.

The grading schemes vary by vendor and by cedant panel, and the exact weighting isn't published the same way across the market. What's consistent is the direction: worse completeness and worse geocoding pull the grade down, and a lower grade pulls the technical price up. Modellers don't have the bandwidth to chase down every missing year-built field by hand, so the grade becomes the proxy for "how much do we trust this schedule."

Where cedants actually lose points

The recurring culprits aren't exotic. Locations added since the last full survey, often from new construction or an acquisition, sit on the schedule with a placeholder address and no verified footprint. Renovations and rebuilds after a loss change the structure but not the SOV line. Secondary locations, the small satellite buildings that never got their own survey, get carried forward year after year with the same stale defaults. None of this is anyone's fault exactly. A schedule of values is a snapshot, usually self-reported, and it goes stale the moment the insured's portfolio changes and nobody re-surveys it.

That's the gap a current built-inventory layer is built to close. Exposure Mapping reads annual high-resolution satellite imagery to produce a consistent view of what's actually standing across a treaty or territory, footprints, counts, and rough form, so a modeller can check a schedule against what's on the ground instead of against last year's survey.

If a chunk of your renewal book is sitting at C or D because the underlying schedules haven't been re-verified in years, it's worth seeing what a fresh built-inventory pass over the territory would show.

Get started

← Back to the blog