Citation Probability: What Makes AI Systems Choose Your Content as a Source
Citation Probability is not a proxy for backlinks. It is a seven-factor SiteNexis analytical model for assessing content signals associated with citation readiness; it is not a provider-published ranking factor or a guarantee of citation.
SiteNexis models citation readiness with seven weighted factors. The model is an internal analytical construct based on observable page and entity signals; it does not expose or reproduce the private selection logic of Google, ChatGPT, Perplexity, or another provider.
The Seven Citation Factors
- Factual Density (20%) — specific, verifiable claims per 100 words. Generic assertions score near zero.
- Claim Specificity (15%) — claims that name entities, dates, percentages, or other verifiable details.
- Primary Entity Authority (15%) — Entity Confidence Score of the page's primary entity.
- Topical Authority Depth (15%) — breadth and depth of supporting content across the domain on this topic.
- Structural Citation Readiness (15%) — structured paragraphs, defined claims, schema markup that makes attribution possible.
- Temporal Freshness (10%) — recency of content, presence of datePublished/dateModified schema.
- Trust Signal Density (10%) — authorship attribution, organisation schema, external validation signals.
Why Factual Density Matters Most
Factual density is the highest-weight factor in the current SiteNexis configuration because the model is intended to distinguish specific, verifiable claims from generic assertions. The score is a diagnostic signal; it does not establish that a provider will cite a page or that a particular threshold predicts citation outcomes.
◆Increase factual density by replacing generic assertions with specific observations, sources, dates, or methodology notes. Label SiteNexis measurements as SiteNexis measurements and do not present them as independent provider evidence.
Structural Citation Readiness
Structural citation readiness is the least obvious factor. AI systems need content structured so that individual claims can be extracted without losing their meaning. A well-structured citable claim has: a clear subject (the entity or concept the claim is about), a specific predicate (what is being asserted about the subject), and evidence or attribution (a data source, observed pattern, or methodology reference). Claims buried in long compound sentences with multiple qualifications are structurally difficult to cite without distortion.
Topical Authority Depth
A single page with high factual density will not score highly on topical authority depth if it is the only page on the domain covering this topic. SiteNexis treats depth across related pages as a supporting diagnostic signal rather than evidence of a universal provider preference. The Perception Graph's topical authority clusters feed this factor: topics that appear as well-supported clusters (multiple pages, multiple entity relationships) score higher in the SiteNexis model than isolated topic pages.