# Only Fan Out > Only Fan Out is a free Chrome extension that shows the fan-out queries, fetched sources and cited sources behind answers in ChatGPT, Claude and Grok. Only Fan Out runs as a Manifest V3 side panel. It supports ChatGPT, Claude and Grok. It has no servers, no accounts and no analytics; all captured data stays in the browser's local storage. It is free to use. The name is also written OnlyFanOut or onlyfanout, and the site is onlyfanout.com. Only Fan Out is an independent tool and is not affiliated with, endorsed by, or connected to OnlyFans or Fenix International. SEO and GEO practitioners use it to see which fan-out queries an AI engine ran, which sources it fetched, and which sources it actually cited. ## Pages - [Home](https://onlyfanout.com/): Product overview, features, screenshots and the latest guides. - [Chrome Web Store listing](https://chromewebstore.google.com/detail/only-fan-out/anoijnnemghjkmlippfljckfidhjafil): Install page (free). - [Blog](https://onlyfanout.com/blog): Guides to how ChatGPT, Claude and Grok search, fetch and cite, and how to use fan-out data for GEO. - [Changelog](https://onlyfanout.com/changelog): What changed in each release, newest first. Version 1.3.0 (submitted 14 September 2026) reads ChatGPT's fan-out queries from the live answer stream, since ChatGPT no longer saves them in the conversation. - [Privacy policy](https://onlyfanout.com/privacy): No servers, no accounts, no analytics. Everything stays in the browser. - [Terms of use](https://onlyfanout.com/terms): License, acceptable use and warranty disclaimer. - [Support](https://onlyfanout.com/support): Common fixes and how to report a bug. ## Guides - [What is query fan-out?](https://onlyfanout.com/what-is-query-fan-out): How ChatGPT, Claude and Grok split one question into many searches, how each engine cites, and why it matters for GEO. - [How ChatGPT, Claude and Grok search the web](https://onlyfanout.com/how-ai-assistants-search): How each assistant decides to search, writes its queries, reads pages and cites sources, from OpenAI, Anthropic and xAI documentation, plus the crawlers that let them read a site. - [How to see what ChatGPT searched before it answered](https://onlyfanout.com/blog/see-what-chatgpt-searched): How to see the search queries ChatGPT ran before it answered, since it stopped saving them on 14 September 2026: the Sources button, Only Fan Out, Chrome DevTools, the OpenAI API and Search Console compared. - [How Only Fan Out works, and why your data stays in your browser](https://onlyfanout.com/blog/how-only-fan-out-works): Where ChatGPT, Claude and Grok run their searches, what they send back to the browser, what the extension reads from those four sites, the requests it makes to re-read a conversation, its Chrome permissions, and why captured data stays in local storage. - [How to find ChatGPT's fan-out queries in Chrome DevTools](https://onlyfanout.com/blog/find-chatgpt-fan-out-queries-devtools): A step-by-step Chrome DevTools method for finding the fan-out queries ChatGPT ran: which request to open, which field to search, what the queries look like, why the search can come up empty, and how it compares with Only Fan Out. - [ChatGPT's four kinds of fan-out query: web, business, image and find](https://onlyfanout.com/blog/chatgpt-fan-out-query-types): The kinds of search line ChatGPT writes (web, business, image and find), what each looks for, the mix in a 17-query turn recorded on 9 September 2026, why a find line is a look-up inside an open page and not a search, and the product, widget and length lines that sit beside them. - [AI citations: what ChatGPT, Claude and Grok read and cite](https://onlyfanout.com/ai-citations): How ChatGPT, Claude and Grok show citations, how many of the pages they read end up cited, and what the Ahrefs and Chen and Koudas studies found about which pages get cited and why brands get named without a link. - [Fetched vs cited sources: why AI answers leave out pages they read](https://onlyfanout.com/blog/fetched-vs-cited-sources): Where ChatGPT, Claude and Grok keep the pages they fetched and the pages they cited, what the Ahrefs, Strauss, Venkit and Peec figures each counted as fetched, why assistants read more than they cite, and a per-turn method for measuring your own domain's cite rate. - [Why ChatGPT read your page but did not cite it](https://onlyfanout.com/blog/fetched-but-not-cited): Why ChatGPT leaves most fetched pages out of its answer: Resoneo's July and August 2026 counts of URLs retrieved, shown, cited and opened (760 of 58,000 opened, 74% of opened pages cited against 7% of pages only retrieved); a 14 September 2026 ChatGPT conversation recorded with Only Fan Out (78 fetched, 15 cited, 7 in both); six reasons with evidence, what to check and what to change; a test of how much of a page's opening fits in a 200-character snippet (5 of 13 of our guides fit their first sentence, 9 of 13 without the date line); why pages from later searches are cited less (Seer, Suganthan and a 9 September 2026 recording); a step-by-step way to find which reason applies; and what GPTBot and structured data do not change. - [GEO with fan-out data: how to use fan-out queries, step by step](https://onlyfanout.com/geo-with-fan-out-data): A six-step GEO workflow built on the searches ChatGPT, Claude and Grok actually run: where fan-out queries come from and how far to trust each source, what each kind of search tells a site owner, how to clean and group a raw query list, what to fix first for each state a turn shows, and how to measure the result. - [How to run a GEO audit with fan-out data, step by step](https://onlyfanout.com/blog/geo-audit-with-fan-out-data): A manual GEO audit built on the searches ChatGPT, Claude and Grok run: crawler and HTML checks before the first prompt, a prompt set by buying stage with three runs each, a two-tab audit sheet for queries and turns, five turn states and four scores by theme, competitors taken from the cited lists, what the report should say, and the limits of a manual audit. - [How to export ChatGPT search queries and sources to CSV](https://onlyfanout.com/blog/export-chatgpt-search-queries): How to get the searches ChatGPT ran and the pages it fetched and cited into a CSV: why exports made from chats saved since 14 September 2026 hold pages and no queries; which routes still give the queries (ChatGPT data export, chat exporters, DevTools, the OpenAI API web search tool, the Only Fan Out side panel); a two-tab sheet layout; row counts from two ChatGPT conversations captured in September 2026 (17 queries from 3 search calls; 86 page rows from 78 fetched and 15 cited, 7 in both); why dividing cited by fetched overstates the rate (19% against 9%); five Google Sheets formulas; and what no export can contain. - [How Claude searches the web: web search, web fetch, places and citations](https://onlyfanout.com/blog/how-claude-searches-the-web): When Claude decides to search, its four tools (web search, web fetch, places search and Bing image results), what Anthropic's docs and published captures say about where its results come from, Anthropic's guidance on searches per request, the rule that Claude only fetches URLs already in the conversation, the blocks and fields of one search turn, and what that means for a site that wants to be read and cited. - [Why ChatGPT's fan-out queries went missing in September 2026](https://onlyfanout.com/blog/chatgpt-fan-out-queries-missing): What ChatGPT removed from saved conversations on 14 September 2026, shown field by field from captures on 9 and 14 September; how the query text is now wiped from the live stream before an answer finishes; what a saved chat still holds (search steps, progress labels, fetched and cited pages); a dated record of the earlier 2026 changes; why a tool shows zero queries, symptom by symptom; and the ways to get the queries back. - [How Grok searches: web searches, page browses and X](https://onlyfanout.com/blog/how-grok-searches): When Grok decides to search, its three kinds of search (web search, page browse and X search) and the four X search calls xAI names, what xAI says about its web search provider, Peec's April 2026 figures for how many queries Grok writes per prompt and how often it uses site: and Reddit searches, the citation fields in one Grok answer, a February 2026 test of the browser identity Grok uses to fetch pages, and what that means for a site that wants to be read and cited. - [Fan-out queries vs keywords: what AI searches add to keyword research](https://onlyfanout.com/blog/fan-out-queries-vs-keywords): How the searches ChatGPT, Claude and Grok write differ from the keywords people type: who writes each, search volume, length and how often fan-out queries repeat (DataForSEO, Semrush and Surfer figures); all 12 searches Claude and Grok wrote for one tuition-centre request in September 2026, word by word; why a keyword tool cannot hold queries built from earlier results; the settings that travel with a fan-out query (search type, recency window, location, result count, operators); and a six-step way to use keywords and fan-out queries together. - [ChatGPT site: searches in fan-out: how often, and which sites](https://onlyfanout.com/blog/chatgpt-site-searches): What a site: search is in ChatGPT and the three places a one-site limit can sit (query text, the domain field of a web search line, a developer's API domain filter); six published counts of how often ChatGPT uses site: (DEJAN, Ahrefs, Promptwatch, MJ Cachón, Peec, Nectiv), from 0.14% to 64%, with each setup and why they disagree; three site: searches in one ChatGPT turn recorded on 14 September 2026; which sites get searched (mostly the brand's own, then Reddit, review sites and public bodies); what to fix on your site; and how Claude and Grok compare. - [Does ranking on Google get you cited by AI? What the studies say](https://onlyfanout.com/blog/rankings-vs-ai-citations): How often AI answers cite pages that rank in Google's top ten, from six studies lined up by engine and yardstick (Ahrefs July 2025: 76.1% of AI Overview citations in the top ten; Ahrefs March 2026: 37.9%; CiteLens: AI Mode 93%, Perplexity 89%, Claude 53%, ChatGPT 30% by domain; Ahrefs August 2025: ChatGPT 8.0%, about 12% across four assistants; Grossman and others; Chen and Koudas); why the figures disagree (engine, top ten or top 100, page or domain, questions, model and date); why none shows that rank causes citation; recordings from September 2026 and a 13 November 2026 Bing check showing the searches an assistant writes move away from the question (9 of 10 results for a tuition request were directories or lists); and a six-step way to track rank for the recorded fan-out queries. - [ChatGPT shopping product cards: what they show beside cited sources](https://onlyfanout.com/blog/chatgpt-shopping-product-cards): What a ChatGPT product card shows (picture, name, one price, one merchant, rating and review count, other shops) set against OpenAI's shopping help page; a real ChatGPT answer to "gifts for men" saved on 9 December 2025 (8 cards carrying 22 offers from 16 merchants, 3 pages cited out of 36 fetched, no shop on a card among the fetched pages); three cards whose price came from the first shop while a cheaper offer carried the Best price tag; ratings that link to a different retailer from the card; the selection order of the cards; whether cards are ads; three ways product data reaches ChatGPT; and how to see cards and cited sources side by side. - [How to track brand mentions in ChatGPT, Claude and Grok answers](https://onlyfanout.com/blog/track-brand-mentions-in-ai-answers): A manual way to track a brand in ChatGPT, Claude and Grok answers that keeps four counts apart (named, cited, searched by name, read but not cited); one question put to Claude on 10 September 2026 and Grok on 11 September 2026, with all 15 tuition centres scored in each; why a mention can be a rule-out; why a domain entry finds citations a brand name misses; seven steps for a tracking round; an eight-column tracking sheet; and what manual tracking cannot tell you. ## Free tools - [Free tools](https://onlyfanout.com/tools): Free tools that run in your browser and read AI search data from files you already have. - [ChatGPT export search reader](https://onlyfanout.com/tools/chatgpt-export-search-reader): Lists every web search saved in a ChatGPT data export, with the pages fetched and cited, as CSV. Runs in your browser and uploads nothing. ## Help - [Help](https://onlyfanout.com/help): Short how-to pages for each Only Fan Out task, from seeing a search to fixing an empty panel. - [How to see what Claude searched](https://onlyfanout.com/help/see-what-claude-searched): Claude's web searches, page fetches and place searches for each answer, with the pages it read and cited. - [How to see what Grok searched](https://onlyfanout.com/help/see-what-grok-searched): Grok's web searches, page browses and X searches for each answer, with the pages and posts it found and cited. - [How to see which X posts Grok used](https://onlyfanout.com/help/see-which-x-posts-grok-used): The X posts Grok's searches brought back for an answer, and how many the answer cited. - [How to see which places Claude used](https://onlyfanout.com/help/see-places-claude-used): The places Claude's place searches returned for a local question, and which ones it mapped or named. - [How to set up the watchlist for your brand and domain](https://onlyfanout.com/help/set-up-the-watchlist): Add brands and domains once, and each captured answer shows whether they were cited, mentioned, fetched or absent. - [How to export Claude and Grok searches](https://onlyfanout.com/help/export-claude-and-grok-searches): Save a Claude or Grok conversation's searches and sources as CSV, Markdown or JSON, or copy one turn's queries. - [How to search your past AI searches](https://onlyfanout.com/help/search-past-ai-searches): Find a past ChatGPT, Claude or Grok conversation by title, prompt or query text in History. - [How to back up your captured data](https://onlyfanout.com/help/back-up-captured-data): Save every captured conversation as one JSON file from History. - [How to pause capture or delete your data](https://onlyfanout.com/help/pause-or-delete-captured-data): Turn capture off without a reload, delete one conversation, or clear everything. - [Why the side panel shows nothing, and how to fix it](https://onlyfanout.com/help/side-panel-shows-nothing): Fixes for a side panel that is empty or misses a search on ChatGPT, Claude or Grok. ## Glossary - [AI search and GEO glossary](https://onlyfanout.com/glossary): Plain definitions of AI search and GEO terms, each with sources. - [What is Google AI Mode?](https://onlyfanout.com/glossary/ai-mode): Google Search's conversational mode that answers harder, multi-part questions with a generated reply and links, running many related searches behind it. - [What are Google AI Overviews?](https://onlyfanout.com/glossary/ai-overviews): Generated summaries that Google Search shows above some results, answering the query in a few lines with links to the pages it drew on. - [What is a canonical URL?](https://onlyfanout.com/glossary/canonical-url): The preferred address for a page when the same or very similar content is reachable at more than one URL, so search engines consolidate signals there. - [What is ChatGPT search?](https://onlyfanout.com/glossary/chatgpt-search): The ChatGPT feature that searches the web when a question needs current information, then answers with inline citations and a list of sources. - [What is ChatGPT-User?](https://onlyfanout.com/glossary/chatgpt-user): OpenAI's user agent for visits a ChatGPT user sets off, such as opening a page to answer a question, rather than automatic crawling. - [What is Claude-SearchBot?](https://onlyfanout.com/glossary/claude-searchbot): Anthropic's crawler for improving the quality of Claude's search results, one of three Anthropic bots that respect robots.txt. - [What is Claude-User?](https://onlyfanout.com/glossary/claude-user): Anthropic's user agent for fetching a web page when a person's question to Claude needs that page. - [What is ClaudeBot?](https://onlyfanout.com/glossary/claudebot): Anthropic's web crawler that collects public content which may be used to train its models. - [What is a context window?](https://onlyfanout.com/glossary/context-window): All the text a language model can take into account at once when it answers, including the conversation, any fetched pages and its own reply. - [What is crawl budget?](https://onlyfanout.com/glossary/crawl-budget): The set of URLs on a site that a search engine can and wants to crawl in a given time, set by how much load the server can take and how much the engine wants the pages. - [What is deep research in ChatGPT?](https://onlyfanout.com/glossary/deep-research): A ChatGPT mode that runs many web searches and page reads over several minutes and writes a report with citations, rather than a quick answer. - [What is E-E-A-T?](https://onlyfanout.com/glossary/e-e-a-t): Short for experience, expertise, authoritativeness and trustworthiness: the qualities Google's search quality raters look for in content, with trust counted as the most important. - [What is an embedding?](https://onlyfanout.com/glossary/embedding): A list of numbers that represents the meaning of a piece of text, so that texts with similar meanings get numbers that sit close together. - [What is a featured snippet?](https://onlyfanout.com/glossary/featured-snippet): A Google result that shows a short extract answering the query first, with the page title and link under it, often at the top of the results. - [What is function calling?](https://onlyfanout.com/glossary/function-calling): A way for a language model to ask the app it runs in to call a named tool, such as a web search, with arguments it chooses, then use the result in its answer. - [What is generative engine optimisation (GEO)?](https://onlyfanout.com/glossary/generative-engine-optimisation): Shaping content so AI answer engines find it, use it and cite it. - [What is Google-Extended?](https://onlyfanout.com/glossary/google-extended): A robots.txt token that tells Google whether content it crawls may be used to train its Gemini models and to ground answers in Gemini Apps. - [What is Googlebot?](https://onlyfanout.com/glossary/googlebot): The name for the web crawlers Google Search uses to fetch pages, in a smartphone and a desktop version that share one robots.txt token. - [What is GPTBot?](https://onlyfanout.com/glossary/gptbot): OpenAI's web crawler that collects content which may be used to train its foundation models. - [What is grounding in AI?](https://onlyfanout.com/glossary/grounding): Tying a language model's answer to sources it retrieved and can point to, such as search results, so the answer rests on more than what the model learned in training. - [What is an AI hallucination?](https://onlyfanout.com/glossary/hallucination): A response from an AI model that presents false or misleading information as if it were fact. - [What is indexing?](https://onlyfanout.com/glossary/indexing): The step where a search engine analyses a page it has crawled and stores it, so it can be returned for searches. - [What is IndexNow?](https://onlyfanout.com/glossary/indexnow): An open protocol that lets a website notify search engines the moment a page is added, updated or deleted, instead of waiting for a crawl. - [What is JSON-LD?](https://onlyfanout.com/glossary/json-ld): A JSON-based format for describing things as linked data, used on web pages to tell machines what the page is about in a block of code. - [What is a knowledge cutoff?](https://onlyfanout.com/glossary/knowledge-cutoff): The date up to which a language model's training gives it reliable knowledge; anything after it the model can only learn by searching or being told. - [What is a large language model (LLM)?](https://onlyfanout.com/glossary/large-language-model): A neural network trained on huge amounts of text, the model behind ChatGPT, Claude and Grok. - [What is llms.txt?](https://onlyfanout.com/glossary/llms-txt): A proposed Markdown file at the root of a website, /llms.txt, that gives language models a short summary of the site and links to its most useful pages. - [What is a long-tail keyword?](https://onlyfanout.com/glossary/long-tail-keyword): A specific, longer search phrase that few people search for on its own but that faces less competition, the opposite of a short, popular head term. - [What is noindex?](https://onlyfanout.com/glossary/noindex): A rule, set in a page's robots meta tag or an X-Robots-Tag HTTP header, that tells search engines not to include the page in their index. - [What is OAI-SearchBot?](https://onlyfanout.com/glossary/oai-searchbot): OpenAI's crawler for ChatGPT search: it lets a site's pages appear in ChatGPT's search answers, and sites that block it are left out. - [What is query expansion?](https://onlyfanout.com/glossary/query-expansion): Rewriting or adding to a search query, with synonyms, related terms or corrected spellings, so it finds more of the relevant results. - [What is retrieval-augmented generation (RAG)?](https://onlyfanout.com/glossary/retrieval-augmented-generation): A model retrieves documents first, then writes its answer from them. - [What is robots.txt?](https://onlyfanout.com/glossary/robots-txt): A plain text file at the root of a website that tells crawlers which parts of the site they may and may not visit. - [What is a search engine results page (SERP)?](https://onlyfanout.com/glossary/search-engine-results-page): The page a search engine shows for a query: links, ads, snippets and AI answers. - [What are server-sent events (SSE)?](https://onlyfanout.com/glossary/server-sent-events): A web standard for streaming updates from a server to a page over one connection. - [What is structured data?](https://onlyfanout.com/glossary/structured-data): Code added to a page that describes its content in a standard vocabulary, usually schema.org, so machines can tell what the page is and what it contains. - [What is a token in AI?](https://onlyfanout.com/glossary/token): The unit of text a language model reads and writes, usually a word or part of a word; in English, one token is about four characters. - [What is a web search tool in an AI model?](https://onlyfanout.com/glossary/web-search-tool): A tool a language model can call to run web searches during an answer, choosing its own queries and citing the pages it uses. - [What is X search in Grok?](https://onlyfanout.com/glossary/x-search): A Grok tool that searches posts on X by keyword, meaning or account, can pull whole threads, and cites the posts it used in the answer. - [What is an XML sitemap?](https://onlyfanout.com/glossary/xml-sitemap): A file, usually sitemap.xml, that lists a site's pages and when each last changed, so search engines can find and crawl them. ## Facts - Supported engines: ChatGPT, Claude and Grok. - Captures per engine: fan-out queries in the order the engine ran them, sources fetched versus sources cited, a brand or domain watchlist, and cross-engine history. - Privacy: no servers, no accounts, no analytics. Captured data stays in chrome.storage.local on the user's device. - Export formats: CSV, Markdown and JSON. - Diagnostics: flags schema drift when an AI site changes its page structure. - Price: free. - Install: https://chromewebstore.google.com/detail/only-fan-out/anoijnnemghjkmlippfljckfidhjafil - Contact: gerald@axecute.space ## Optional - [Full text](https://onlyfanout.com/llms-full.txt): The home page, the changelog, the main guides and every blog post as plain Markdown, in one file.