Evidence first, verdicts second

What we do

We run a deterministic audit against a website and report what we found, what it means, and what to do about it. Every check has a stable identifier, a captured piece of evidence, and an explanation of why it matters. Nothing is asserted without showing the basis for it.

How the score is produced

The audit is rule-based. It reads the pages your site actually serves and applies documented checks to them — crawler directives, structured data, page structure, trust signals, and dates.

What we do not claim

• Our score is produced by us. It is not used by OpenAI, Google, Anthropic, Perplexity, Microsoft, or any other platform, and it is not an official ranking factor anywhere.
• Improving it does not guarantee that a site appears in, or is cited by, any AI answer. No technique does. What technical work does is remove reasons for exclusion.
• Some checks are heuristic — they infer a signal by pattern matching and can miss one expressed unusually. Those are labelled in the report, with a note on how each can be wrong.

What is next

Media Trust is in development: an evidence-based image report covering Content Credentials, metadata, and durable fingerprints. It follows the same principle — show the evidence, state the uncertainty, and avoid turning an inference into an accusation.

Current status

The website audit is in free beta. The methodology is versioned and published, and reported figures describe audit completion and coverage rather than finding accuracy, which is still being reviewed against a labelled benchmark set. We would rather say that plainly than imply a precision we have not yet demonstrated.