Example Prompts
- “Do AI assistants recommend [brand]?”
- “GEO audit for [brand]: 8 prompts, all three engines”
- “Who wins [category] in AI answers, and why?”
- “Which sources do AI engines assemble [category] answers from?”
Audit Methodology
1
Build the prompt panel
3-8 questions buyers actually ask assistants, derived from the product’s category, personas, and competitor set (community mining via Xpoz helps surface real phrasings).
2
Trace every prompt
Each prompt runs through Claude, ChatGPT, and Gemini with live web search on, capturing answers, the engines’ search queries, and cited URLs (the ai-answer-trace skill provides the tracing).
3
Judge visibility per prompt
Present (recommended, listed, mentioned, or absent), cited (brand-owned URL in the citations), who wins instead and via which URLs, and stability across samples.
4
Report
Per-prompt verdict table, cited-domain rankings with surface types, brand citation paths, lost prompts diagnosed, and recommended next steps.

