How do you measure AI visibility accurately?
Measure the share of sampled AI answers in which your brand appears, across many prompts run many times — not a single scan, and not an "AI ranking position". AI assistants answer the same question differently on different runs, so one scan is a snapshot rather than a measurement. Rand Fishkin of SparkToro, starting from open scepticism, concluded from a 2,961-prompt-run study that visibility percentage across repeated runs is a legitimate trackable metric, while any tool reporting a ranking position in AI is, in his words, "full of baloney".
Report mentions and citations separately. They are different phenomena: on Gemini the overlap between brands mentioned and domains cited can be as low as 30%, and ChatGPT mentions brands roughly 3.2 times more often than it cites them. A tool that collapses the two hides which problem you actually have.
Attach a confidence interval and a run count to every number. A rate from 3 checks and a rate from 300 look identical on a dashboard and mean completely different things. Wilson intervals are appropriate here because sample sizes are usually small.
Ask any vendor to show their math: how many prompts, how many runs each, and whether you can read the raw answers behind the number. Most published citation statistics in this category do not disclose their sampling method at all.
Sources
Last reviewed 2026-08-06. We revisit these when the underlying research changes.
Related
See where you actually stand
Run a free check across ChatGPT, Perplexity, Gemini, Claude and Google AI Mode.
Check your AI visibility