What is an AI search visibility score, and how is it measured?
It's whether an AI assistant names you back when a buyer asks it a question you'd want to answer — a different signal from a search ranking, and one most companies have never checked. A page can rank on Google and still never get spoken aloud in a ChatGPT, Claude or Perplexity answer, because the assistant is synthesizing a response from whichever sources it judges citable, not listing URLs. AgentKit Works measures this directly in its MCP Visibility Index: ten buyer-intent queries run against live AI answer engines for each of 111 agent and MCP tool vendors, recording whether the vendor got named and who was named instead. The result as of this run: 59 of 111 vendors — more than half — were never cited once, and the median vendor's score is zero. A visibility score isn't a judgment on whether the product is good; it's a measurement of whether the content around it is structured for an assistant to quote.
Related questions
How do I check my own company's AI search visibility score?
Ask the AI assistants directly — type the buyer questions a prospect would actually ask ("best X for Y", "where to buy X") into ChatGPT, Claude and Perplexity, and note whether your name shows up and who gets named instead. That's the exact method behind AgentKit Works' MCP Visibility Index; any vendor can run a smaller version of it by hand before paying for anything.
Is an AI visibility score the same thing as a Google ranking?
No — a ranking is a position for a URL in a results list; a visibility score is whether your name gets spoken inside an assistant's synthesized answer. AgentKit Works tracks both separately because they move independently: a page can hold a decent Google position and still never get named in an AI answer, since the assistant is choosing which sources to cite, not which links to list.
Why would a company rank on Google but still score zero on AI visibility?
Because the two systems extract different things. A search engine indexes a page; an answer engine needs an extractable claim it can quote and attribute, which means FAQ schema, a direct answer near the top of the page, and llms.txt pointing it to the right content. AgentKit Works' AI Search Visibility Kit (the self-serve version of the Visibility Index methodology) builds pages around exactly that extractable structure rather than classic ranking signals alone.
What counts as a 'citation' in the Visibility Index, specifically?
The vendor's own name appearing in the assistant's answer text for a buyer-intent query — not a link in a sources list, not a mention buried in a comparison table the assistant didn't surface. AgentKit Works records a strict yes/no per query (plus who got named instead when the answer was a no) rather than a fuzzy relevance score, because a buyer reading the answer either heard your name or didn't.
Does a zero visibility score mean the product itself is bad?
No — it means the surrounding content has never given an assistant anything to quote, which is a fixable content problem, not a verdict on the product. Most of the 59 zero-scored vendors in AgentKit Works' Visibility Index sell real, working tools; they're invisible to this specific channel because nothing on their site is structured as a citable answer.
Can AgentKit Works fix a low AI visibility score, or is this just a diagnostic?
Both — the Visibility Index is the free diagnostic, and the AI Search Visibility Kit ($79) is the same checklist as a self-serve pipeline: FAQ schema, llms.txt, citation hooks and weekly citation tracking, chained with the keyword-to-page pipeline that produces the content in the first place. AgentKit Works also runs the same checklist as a $399 done-for-you engagement for anyone who wants it applied rather than run themselves.
THE TOOL FOR THIS JOB
AI Search Visibility Kit
Get cited by AI assistants — 13 chained skills for AEO + classic SEO, the DIY tier of our $399 service.
13 skills · 64 failure modes · 3 templates
Runs unattended
View spec