Because each answer is assembled from the pages fetched for that asking, and the searches vary. The core of the list, the companies on most of the pages, is stable; the tail changes. That is why one asking proves nothing and fifteen questions across four engines, repeated monthly on the same questions, give a number you can act on.
What is stable
The companies that appear on most of the sources: the review site's top five, the names in every roundup. They are named in nearly every asking. A company in that core has high share of voice; a company in the tail is named one time in three.
What the report does about it
It asks each question once per engine and treats the answer as a sample; it reports share of voice across all answers, not any one; and it keeps the questions so the next month's check is on the same ones. Movement is measured against the same questions, not against a different asking.
The term behind it: Answer engine
A system that answers a question with a written answer rather than a list of links: ChatGPT with web search, Perplexity, Google's AI Overviews and AI Mode, Gemini with search grounding.
The report scores four: ChatGPT with web search, Perplexity, Gemini with Google Search grounding, and Google AI Overviews. Each is asked every question with the buyer's country set and scored separately, so the split between engines is visible. Answer engine, in full.
Read next
How the AI visibility score works · Share of voice · Posture
Ask it about your own company
Your website, region and work email. The score appears in about three minutes, and the report is a link you keep.