LLM visibility · reference

LLM visibility: what to measure and inspect

LLM visibility can describe how a brand or page appears in language-model answer experiences. The useful starting point is a precise evidence definition, not an invented universal score.

Start free trial Read the guide

Direct answer

Keep the search question explicit.

Define LLM visibility by provider, target, operation, and returned fields. A returned mention or source is evidence for that context, not proof of comprehensive visibility or future performance.

Reference guide

Concepts and evidence boundaries.

Name the evidence

A useful observation may include provider or model context, a target, sources, search results, brands, or AI Search Volume where the operation returns those fields.

Choose occurrence, comparison, or change

Use mention evidence for an occurrence question, a consistent target set for comparison, and historical, delta, or new-and-lost evidence only when the question is about change.

Avoid the universal-score trap

Different providers and operations answer different questions. Do not turn one returned result into a comprehensive score, share-of-voice calculation, or market ranking.

Keep selection constraints visible

Supported product evidence is conditional on its provider and operation contract; ChatGPT selection, for example, is constrained to United States and English.

Separate availability from visibility

Crawler access, indexing eligibility, and returned source evidence answer different questions. Record each layer separately instead of collapsing them into one visibility score.

Primary sources