The AI Visibility Score: What It Measures and Why Each Component Matters

The AI Visibility Score is SiteNexis's composite measure of how visible content is to AI retrieval and citation systems. This article documents each component, the reasoning behind the formula weights, and what improving each component actually produces in practice.

The Six Components and Their Weights

The AI Visibility Score is calculated as: Machine Readability Score × 0.15 + Entity Confidence Score × 0.20 + Retrieval Readiness Score × 0.20 + Citation Probability Score × 0.20 + Semantic Trust Score × 0.15 + Schema Completeness Score × 0.10. The weights reflect both the relative importance of each dimension to AI citation probability and the relative leverageability of each dimension — components that have higher impact on citation outcomes and are more directly improvable receive higher weights.

Machine Readability Score (15%)

Measures extraction fidelity: how much meaning survives the AI extraction pipeline from raw HTML to usable semantic chunk. Weighted at 15% because it is a necessary condition but not the primary differentiator between high and low citation rates among sites that have reached basic machine readability thresholds. Sites below the 50-point threshold on machine readability typically see outsized improvements from addressing readability issues because they are at the stage where the fundamental extraction quality is limiting everything else.

Entity Confidence Score (20%) — Highest Weight Among Semantic Dimensions

Measures entity clarity, consistency, coverage, and disambiguation. Weighted at 20% because entity ambiguity is the most common cause of AI invisibility in sites that are otherwise technically sound — and because entity clarity improvements propagate across all pages on the domain simultaneously, making them high-leverage investments. A 10-point improvement in Entity Confidence typically produces larger improvements in overall AI citation rate than a 10-point improvement in any other dimension.

Retrieval Readiness and Citation Probability (20% each)

Retrieval Readiness measures whether content is structured for AI extraction across six query type models. Citation Probability models the likelihood of citation selection based on factual density, claim specificity, and authority signals. Both are weighted at 20% because they address the most directly improvable dimensions of content performance — both can be improved through specific, targeted content changes without requiring infrastructure changes.

Why Schema Completeness Is Weighted Lowest (10%)

Schema Completeness is weighted at 10% not because it is unimportant, but because most sites with reasonable technical SEO practice have sufficient schema coverage for the score to not be the primary limiting factor. The score weight reflects the marginal impact of schema improvements once a baseline level is achieved. For sites with no schema at all, improving schema completeness produces outsized results relative to the 10% weight — because the baseline was so low that even reaching a moderate level produces significant score improvement across multiple dimensions simultaneously.