- The AI Visibility Evidence Model is a reference model that orders publisher-side factors behind AI visibility by strength of evidence, with five factors: Topical Relevance, Machine Access, Entity Consistency, Extractability, and Independent Corroboration.
- The model is descriptive, not prescriptive; it reports evidence grades (A-D) based on peer-reviewed research, official documentation, and controlled preprints, without promising results.
- Topical Relevance (Grade A) is the strongest content-side driver, as controlled studies show topic match dominates citation, while off-topic content is rarely cited.
- Machine Access (Grade B) is a gate: crawlers must access content for it to be used in AI answers, per platform documentation from OpenAI and Google.
- Entity Consistency (Grade C) reduces disambiguation errors but lacks causal proof as a citation driver; structured data is not required for AI features.
- Extractability (Grade B/C) shows position effects and evidence density matter, but answer-first page structure is correlational and not a guaranteed lever.
- Independent Corroboration (Grade B/C) involves third-party mentions influencing parametric knowledge, but independence as a causal factor is unsupported.
- Several factors receive Grade D, including llms.txt as a visibility signal, content rewriting tricks, schema as a citation switch, and manufactured mentions.
- Visibility is not the same as citation fidelity; audits show AI systems often provide incorrect source attributions, so tracking accuracy separately is essential.
- The model includes a method note, limits, and a source register with 15 cited references, emphasizing non-deterministic systems and the need for updated evidence.