How CoverageCard grades a policy

A CoverageCard grade is our own assessment of how well a policy protects the person who holds it — not an official rating, and not a judgement about the insurer that wrote it.

Every scan runs the same way. An automated pipeline reads the declarations page, pulls out the coverages, limits and deductibles that are actually printed on it, and compares them against a fixed rubric. The letter is the output of that comparison. This page explains the rubric in plain English so the grade is something you can argue with rather than something you have to take on faith.

What the letters mean

Grades run from S at the top through A, B, C, D to F, with plus and minus steps in between.

• S — every major protection present, limits strong and consistent, nothing important declined, and no gaps found. It is deliberately rare.
• A — high limits and nothing important missing. Any gaps found are conveniences rather than exposures.
• B — solid on the fundamentals, with one or two everyday situations that would come out of pocket.
• C — the foundation is there, but real exposure sits alongside it.
• D — one step above the legal floor, with the pieces that matter most missing.
• F — state-minimum coverage and little else. Legal to drive; thin against anything serious.

For car insurance the bands are anchored on the bodily injury liability limits, because that is the number that decides what happens in a serious at-fault crash. For home insurance they are anchored on the settlement basis for the dwelling — replacement cost against actual cash value — and on the personal liability limit.

Severity, and why one gap can cap a grade

Every gap the analysis finds is rated high, medium or low severity. Severity is about consequence, not about how unusual something is.

A single high-severity gap caps the grade at B−, no matter how strong the rest of the policy looks. Uninsured motorist coverage declined, liability sitting at the state minimum, a financed vehicle carrying no comprehensive or collision, a dwelling settling on actual cash value — each of these can undo an otherwise good policy in one claim, so the rubric does not let strong numbers elsewhere average them away.

The cap is applied by our own code after the analysis is written, not by the language model. When it fires, the scan record keeps a note of what the raw grade was before the cap — which is also how we monitor whether the rubric and the model are drifting apart.

Value-aware calls

A rule applied without judgement produces silly results. The clearest example is an old vehicle carrying liability only.

On a twenty-year-old truck with no lienholder, dropping comprehensive and collision is frequently a deliberate and sensible economic decision — the coverage can cost more over a few years than the vehicle would pay out. The rubric treats that as a low-severity observation worth confirming, and low-severity items never cap a grade. The same missing coverage on a financed vehicle is treated as high severity, because a total loss there leaves a loan with nothing behind it.

Bundles get two grades, then one

When a scan covers an auto and home bundle, each policy is graded independently against its own rubric. Our code then blends the two sub-grades evenly into the overall letter, and where a blend lands exactly between two letters it rounds down rather than up. Both sub-grades stay visible on the card, because a strong home policy and a thin auto policy is a meaningfully different situation from two average ones.

What a grade is not

• It is not an official or industry rating. It is CoverageCard's opinion, produced by our own methodology, and it carries no endorsement from any insurer or regulator.
• It is not a judgement about your insurer. The analysis reads one document. It says nothing about a company's financial strength, claims handling or service.
• It is not a price score. A cheap policy and an expensive one can earn the same letter. The rubric grades protection, not cost.
• It is not advice. A described gap means our methodology did not find that coverage documented in what you uploaded — nothing more.
• It can be wrong. The analysis is automated and reads only what is legible on the pages you provide. Endorsements living elsewhere in your policy will not appear.

Terms used on this page are explained in the Insurance Glossary. For what coverage typically costs where you live, see The Going Rate. More about who runs CoverageCard is on the About page.