Two of SuperSchema’s scores are calculated from inputs SuperSchema can inspect: the per-schema quality score (how good one piece of JSON-LD is) and the site-wide SuperSchema Score (how ready your whole site is for AI). A third, the Citation Score, is not calculated from your markup at all; it is measured by asking AI assistants real questions and counting how often they cite you. This article is about the two calculated ones and what moves them.
The site SuperSchema Score #
The SuperSchema Score is a weighted average of five category scores, each from 0 to 100. Each category is multiplied by its weight, and the results are summed into a single 0 to 100 number. The weights are fixed:
| Category | Weight | What it rewards |
|---|---|---|
| Access | 15% | AI crawlers can reach and render your pages. |
| Content | 25% | Substantial, well-written content to learn from. |
| Structure | 15% | A clean, logical HTML hierarchy. |
| Schema | 25% | Valid, relevant JSON-LD with complete fields. |
| Alternate | 20% | Accessible Markdown and plain-text versions. |
Because it is weighted, the same raw improvement counts more in a heavier category. Lifting Content or Schema by ten points moves your overall score more than lifting Access or Structure by the same amount. Content and Schema carry the most weight (25% each) for a reason: AI engines lean hardest on real substance they can quote and on explicit structured data that tells them what a page is. A page can be technically perfect and still score poorly if it is thin or has no schema.
The per-schema quality score #
When you generate a single schema, it gets its own quality score from 0 to 100, shown as a "Schema Quality Overview". This is a completely separate calculation from the site score, and it looks at just that one piece of JSON-LD. It has two headline numbers:
- Completeness
- Whether the right properties are present. It weighs required properties most heavily, then recommended properties and advanced AEO features, then general content quality. Missing the basics costs the most.
- Quality Depth
- Whether those properties are implemented richly rather than as bare strings: structured objects, optimized descriptions, entity disambiguation (@id, sameAs), media richness, and semantic depth. This is what AI refinement improves most.
Under the hood the completeness side blends four factors: required properties (weighted most), recommended properties, advanced AEO features, and content quality, with a small bonus or penalty for schema.org compliance. Adding a missing required field moves the number more than polishing an optional one, which is why the fastest gains come from filling gaps first, then enriching.
How the two relate #
The schema quality score grades one schema; the Schema category inside the SuperSchema Score grades your site’s structured data across the pages that were scanned. Generating strong schemas on your important pages raises the Schema category, which is 25% of the site score. So the two scores are linked, but improving one schema is a step toward a better site score, not the same thing as it.