Query Fan-Out Revealer for ChatGPT, Claude & Perplexity

Reveals the fan-out queries ChatGPT, Claude and Perplexity run, which sources they cite, and how often your brand appears.

When you ask ChatGPT, Claude or Perplexity a question, it rarely searches for what you typed. It quietly expands your prompt into several different searches, reads the results, and writes an answer from a handful of them. Query Fan-Out Revealer shows you that hidden step, as it happens, in your own sessions. WHAT IT SHOWS • The real fan-out queries — the searches the assistant actually ran, listed in order • Every source it retrieved, grouped by domain • Which of those sources it cited, and which it read and passed over • How often your own domain is retrieved, cited, and named in the answer text • Share of voice against a competitor set you define OBSERVED, NOT SIMULATED Most tools in this space guess what an AI system would probably search for, or run their own prompts on a schedule and report back. This one reads the actual response stream in your own browser, in your own logged-in session. What you see is what happened to you. RETRIEVED IS NOT THE SAME AS CITED Assistants read far more pages than they quote, and the gap is the useful part. Enter your domain and every answer tells you which side you landed on: retrieved and cited, retrieved but passed over, or never found at all. Three different problems, three different fixes, never blurred together. METRICS THAT SHOW THEIR WORK Retrieval rate, citation rate, retrieval-to-citation rate and mention rate. Hover any one and it shows both the plain-language definition and the arithmetic behind the number currently on screen, so you can check a figure before you report it. Prompts that already name your brand are excluded from mention rate and flagged on the card. An answer echoing your own prompt is not organic visibility. EXPORT One click to CSV or JSON. Fan-out queries, tool calls, page fetches and follow-up suggestions stay in separate columns, so a count is never inflated by something that wasn't a search. Every source is marked cited or retrieved_only. PRIVATE BY DESIGN No server, no account, no analytics. The extension reads a copy of the response your browser already received, stores it on your computer, and sends nothing anywhere. It only runs on chatgpt.com, claude.ai and perplexity.ai. Clear all deletes everything permanently. Measuring brand mentions requires storing the answer text locally. You can switch that off in the panel — every other metric keeps working. Built for SEO, GEO and AEO practitioners, content strategists and anyone researching how AI search actually retrieves and ranks information. Not affiliated with OpenAI, Anthropic or Perplexity AI.