Why “AI search visibility” is more precise than “LLM visibility”

A language model and an AI search product are not the same thing. Some products search the live web, some retrieve from selected sources, and some answer without performing a new search. The phrase “LLM visibility” hides those differences and can make an observation sound more universal than it is.

AI search visibility focuses on a public discovery experience that can be observed. It does not claim access to a model’s internal reasoning or training data.

A mention is not enough

An organisation may be named but described inaccurately, placed in the wrong category, or supported by an unreliable source. A useful review therefore examines:

• whether the organisation appears;

• how it is described;

• whether the description is current and accurate;

• which products, services, people, sectors, or locations are associated with it;

• whether sources are shown;

• which sources are used;

• whether the answer changes when the wording changes; and

• whether the experience produces a visible referral or follow-up action.

For an investor firm, accuracy may concern its sectors, stages, geographies, partners, portfolio, or stated approach. Frequency without accuracy is not a positive result.

How it should be measured

A responsible review records the exact question, product, date, location, account state, answer, cited sources, and interpretation. The same prompt should be repeated under agreed conditions over time. Changes should be reported as observations, not as proof that one content change caused them.

There is no responsible universal AI visibility score. A fixed prompt set can show movement within its defined sample, but it cannot represent every possible user, question, or system.