The Machine Trust Score: Trust Is Not Visibility

AI visibility measures whether your content can be found and retrieved. Machine trust measures whether an AI system would confidence-weight your content as a reliable source once retrieved. These are fundamentally different problems. A site can be perfectly extractable, chunked cleanly, entity-rich, and schema-complete — and still be low-trust because its entity claims contradict each other across pages, or its schema markup over-claims attributes not evidenced in body text. Trust is not a byproduct of visibility. It is a separate signal dimension.

Five Components of Machine Trust

The Machine Trust Score is composed of five independently measurable sub-scores. Entity Credibility Consistency (30% weight) measures whether entity attributes are described identically across all pages, schema, and metadata. Schema Trust Alignment (20%) measures whether schema markup accurately describes content without embellishment. External Validation Depth (25%) measures how many entity claims are verifiable through independent external sources via sameAs links. Contradiction Absence (15%) detects cross-page factual conflicts. Trust Degradation Resistance (10%) measures resilience against trust decay signals.

Entity Credibility: The Foundation

Entity credibility is the highest-weighted component because it is the most fundamental. If your Organization schema says you were founded in 2018 but your About page says 2019, and your LinkedIn sameAs link shows 2017, you have a three-way contradiction on a single attribute. AI systems cannot resolve this — they can only reduce confidence. Every inconsistency costs trust. The Entity Credibility Consistency sub-score checks: name consistency, founding date, description, location, category, and key attributes across every page and every schema instance on the domain.

Schema Trust Alignment: No Over-Claiming

Schema markup is a trust signal only when it accurately describes page content. Schema that claims attributes not evidenced in the body text is over-claiming — and AI systems are increasingly capable of detecting this. If your schema declares aggregateRating with 4.8 stars but no review content is visible on the page, that is a schema trust alignment failure. If your author schema names a person who is never mentioned in the article text, that is a misalignment. Every schema claim must be verifiable from the page content itself.

Trust Degradation: Formed and Then Damaged

Trust degradation is distinct from low trust. Low trust means trust was never established. Degradation means trust was formed and then damaged — which is worse from an AI system perspective because it suggests unreliability over time. Degradation signals include: pages that previously had schema now lacking it, entity attributes that changed between audits without explanation, external validation sources that previously resolved but now return 404, and sudden entity changes with no contextual update. The Trust Degradation Resistance sub-score requires audit history to compute — on first audit, it returns baseline values.