How Robot Visible measures AI visibility

Two related measurements

Robot Visible separates website readiness from observed answer visibility. A website scan evaluates the public page and supporting site signals that affect access, interpretation, and trust. Citation monitoring runs selected customer questions and records whether the resulting answer mentions, recommends, or cites the monitored brand and domain.

Neither measurement is a promise of ranking or citation. A readiness check identifies observable conditions on the site; an answer check records what happened for a specified question, provider, configuration, and time.

Evidence retained for an answer check

• The exact customer question and its classified intent.
• The provider, run time, result state, and captured answer evidence available from that provider.
• Whether the brand was mentioned, recommended, or supported by a customer-domain citation.
• The cited URLs and named competitors found in the answer.
• Where available, a crawl-based comparison of a competitor's cited page with the monitored customer page.

Core definitions

How website checks are interpreted

The scanner fetches publicly available URLs and evaluates observable signals such as status, robots directives, canonicals, sitemap discovery, metadata, headings, visible text, structured data, internal links, supporting-page coverage, and trust context. Checks are grouped into six categories and weighted into a summary score.

Thresholds are diagnostic guardrails, not universal ranking factors. For example, a low word count can reveal a page with too little context, but a longer page is not automatically better. The underlying check, page purpose, and competing evidence must be reviewed together.

Known limitations

• Generated answers can vary by time, location, provider, model, search partner, personalization, and query rewriting.
• A successful crawl by Robot Visible does not prove that every search or AI crawler receives the same response through a CDN or firewall.
• A mention or citation can change without a corresponding website change, so several comparable runs provide stronger evidence than one result.
• Competitor-page comparisons currently describe observable differences; they do not prove which difference caused the citation.
• Referral traffic understates visibility when users consume an answer without clicking, while mentions alone do not prove commercial impact.

How to use the evidence responsibly

Treat recommendations as testable hypotheses. Record the intended question and page, publish one coherent improvement, preserve the deployment date, and observe several subsequent runs. Where a result improves, inspect the cited URL and passage before attributing the change.

Robot Visible should be used alongside first-party search-console data, analytics, crawler logs, customer research, and editorial judgment. The strongest conclusion is one supported by several independent forms of evidence.

Research and privacy principles

• Publish aggregate findings only when the sample and inclusion criteria are stated clearly.
• Do not expose private account data, unpublished queries, or identifiable customer results without permission.
• Use minimum cohort sizes and suppress small segments that could identify an individual site.
• Separate measured observations from interpretations and disclose material limitations.
• Record the collection period and update methodology when checks, providers, or definitions change.