Glossary · Glossary

Visibility Score

A visibility score is a vendor-reported number that summarizes AI answer presence. It is a claim about visibility, not the evidence behind it.

A visibility score is a single number a tool reports to summarize how often and how prominently a brand appears in AI-generated answers. Vendors compute it differently: mention rates, positions, sentiment weightings, or blends the vendor does not publish. The score is a summary of a measurement run, not the measurement itself.

Why it matters

Scores are convenient precisely because they compress. Leadership can track one number, a dashboard can chart it, and a vendor can demo it. The operator question is what the number compresses: which prompts, which engines, which dates, and what happened inside the answers.

A defensible score sits on retained records — the prompt wording, the full answer text, the citations, the competitors in the same answer, and the collection date. When those records exist, a score is a useful index into them. When they do not, the score is a claim nobody can audit, and “the score went up” cannot survive the follow-up question of which answers changed.

How it differs

AI visibility is the property being measured: whether and how well the brand appears in answers. A visibility score is a vendor’s numeric summary of that property on one run. The property exists without the tool; the score does not.

AI share of voice is a comparative benchmark — the brand’s slice of mentions versus competitors. Many vendors fold share of voice into a blended score, which is one more reason the blend needs explaining.

An answer monitoring record is the underlying evidence layer. A score summarizes monitoring runs; monitoring does not reduce to the score.

Classic rank positions are external search results. A visibility score usually describes presence inside generated answers, which mix retrieval, synthesis, and framing rather than a ranked list.

Example

A tool reports 62 for the brand this week, up from 55. The team opens the record and finds the lift came from three category prompts where a recent comparison page entered the citations, while a fourth prompt flipped to a competitor. The score was the pointer; the records were the finding. Reporting says “three prompts improved after the page shipped,” not “score up seven.”

Counterexample

A quarterly deck reports the same 62-to-55 move with no retained answers, no prompt list, and no citation records. Nobody can say which engines, which prompts, or which competitors moved. The number cannot be checked, reproduced, or defended, and the next quarter’s dip will be equally unexplained.

Common misunderstanding

A higher score is not automatically a business result, and a vendor’s score is not comparable across vendors — each blend weights prompts, engines, sentiment, and position its own way. A score that only ever rises after any change is a report artifact, not evidence that the change worked. Treat the score as an index into evidence, and ask any demo for the underlying answer and source record behind it.