Citability

When an AI engine retrieves pages to answer a question, it works passage by passage: chunks of text are ranked for relevance and the best are quoted or summarized with attribution. Citable content survives this pipeline. Its passages stand alone without needing surrounding context, state claims explicitly rather than implying them, include concrete numbers, dates, and named entities, and sit under headings that match the questions being asked, behavior closely related to passage ranking in modern search.

Vague, meandering, or purely promotional text fails at every stage: it ranks poorly as a passage, extracts badly, and gives the model nothing factual to attribute.

Practical citability assessments score pages on several axes. Extractability: are there 40-to-80-word passages that fully answer a question? Factual density: does the page state verifiable specifics, statistics, prices, dates, comparisons? Structure: question-led headings, lists, tables, and structured data that machines parse cleanly. Freshness: visible update dates and current information, since engines discount stale sources. Accessibility: content reachable by AI crawlers and present in raw HTML rather than locked behind JavaScript rendering.

Authority signals, author credentials, original research, third-party corroboration, then determine which of two equally extractable pages the engine prefers.

Rewrite key pages answer-first: each section opens with the complete answer, then elaborates. Convert implied knowledge into stated facts, replace "affordable pricing" with the actual number. Add original data where possible, because being the primary source for a statistic makes you the mandatory citation. Keep comparison content ruthlessly current, and remove crawler blocks.

Because citability is page-specific, audit it per URL: Geonimo's page audits score individual pages on AI citability, authority, and readability, flagging exactly which passages prevent an AI citation from happening.