Measuring AI visibility: the new share of shelf

You can't manage what you can't see, and AI answers are slippery — they vary by phrasing, they're personalized, and there's no ranking report. So a new measurement discipline has emerged: a brand-visibility pipeline that treats the AI itself as something to survey.

The method mirrors share of search, adapted for answers:

1. Define a prompt set — the real questions shoppers would ask an AI in your category ("best budget earbuds for running," "wireless earbuds that survive sweat"). This is the answer-era version of a keyword basket. 2. Query the engines repeatedly — run those prompts across ChatGPT, Gemini, Perplexity, Rufus, etc., ideally multiple times (answers vary). 3. Parse the responses — for each answer, detect: Was our brand mentioned? In what position? Described positively or negatively? Were competitors named instead? What sources were cited? 4. Score & trend — compute a share-of-voice in AI answers, by prompt, engine, and over time.

Prompt: "best sweat-proof running earbuds under $150"
  ChatGPT   → mentions us #2 of 4  ✔  (cites our PDP + a review site)
  Gemini    → competitors only     ✘  (gap!)
  Perplexity→ mentions us #1        ✔

That output is gold: it shows exactly where you're absent, which competitors own the answer, and which sources the AI trusts — so you know whether to strengthen content, chase reviews, or earn third-party mentions. It's share of shelf reinvented for the answer era, and the direct feedback gauge for all the GEO work in this series.