What does this AI visibility score measure?
It summarizes observed answers from a defined prompt test. The composite weights brand visibility at 50%, prominence at 20%, claimed-domain citations at 20%, and factual accuracy at 10%.
Free transparent calculator
AI visibility score calculator
Turn real prompt-test counts into a transparent composite while keeping visibility, prominence, citations, accuracy, and coverage separate.
Published:
Observed-answer inputs
Enter counts from a real prompt test
Answers successfully collected with the required grounding.
Count the brand only when it appears in the answer itself.
Answers where the brand leads or appears in the primary shortlist.
Eligible answers linking to the brand's canonical domain.
Mentioning answers with materially correct positioning and facts.
Transparent composite
32/ 100
Composite = 50% visibility + 20% prominence + 20% claimed-domain citation rate + 10% factual accuracy. Keep the components visible; the composite alone cannot explain what changed.
- Visibility
- 25%
- Prominence
- 40%
- Owned citations
- 12%
- Accuracy
- 88%
mentions ÷ eligible answers
prominent ÷ mentioning answers
claimed-domain citations ÷ eligible answers
accurate ÷ mentioning answers
Interpretation guardrail
A score is a summary of a test—not a property of the brand
The same brand can score differently when the question set, providers, locale, web-search behavior, date, or eligibility rules change. Save those conditions beside every score.
Compare reruns only against the same frozen benchmark. Read the component rates before acting: a citation problem requires a different response from an accuracy or discovery problem.
Common questions
Use the resource without overstating the result
Does the calculator scan a website?
No. Website readiness and observed answer visibility are different measurements. Enter counts from real, time-stamped AI answers rather than inferring visibility from schema, word count, or other proxies.
What counts as an eligible answer?
An eligible answer completed under the defined protocol and met any required grounding or source conditions. Failures and unsupported runs belong in coverage, not silently in the miss denominator.
Is the composite comparable across tools?
Only when the tools use the same questions, providers, collection conditions, eligibility rules, and formula. Otherwise similar-looking scores can measure materially different things.
Evidence before conclusions
See a complete benchmark before running your own
Review the questions, stored answers, citations, competitors, coverage, and five prioritized actions in the sample report.