How ChatGPT Search Chooses Which Websites to Cite

ChatGPT Search is a search feature integrated into ChatGPT that retrieves current web content to supplement AI-generated responses. When a user asks a question that benefits from up-to-date information, ChatGPT Search queries the web, reads relevant pages, and incorporates that content into its answer - citing the sources it used.

For website owners, ChatGPT Search represents a growing source of potential visibility. As more people use ChatGPT as their first stop for answers rather than a traditional search engine, appearing as a cited source in ChatGPT responses is an increasingly relevant goal.

This guide explains how ChatGPT Search works, what affects source selection, and how to improve your chances of being cited.

For the broader GEO context, read What is Generative Engine Optimisation (GEO)?

How ChatGPT Search Works

ChatGPT Search is powered by a connection to the Bing web index. When ChatGPT determines that a query would benefit from current web information, it retrieves a set of relevant pages from the Bing index, reads them, and incorporates relevant content into its response.

The citations appear as numbered references within the response text, and source links are typically shown alongside or beneath the answer. Users can click through to the original pages.

ChatGPT determines whether to use search based on the query type. Questions about current events, recent data, specific facts, and how-to tasks frequently trigger a web search. Abstract or creative requests typically do not.

This architecture has two practical implications. First, being indexed by Bing is a prerequisite for ChatGPT Search to access your content. Second, the same quality signals that Bing uses to rank pages for traditional search results influence which pages are retrieved and prioritised for ChatGPT Search responses.

Being Indexed by Bing

Many website owners focus exclusively on Google Search Console and Google indexing, overlooking Bing. Since ChatGPT Search draws from the Bing index, Bing indexing is a direct prerequisite for ChatGPT Search visibility.

To check and manage Bing indexing:

Bing Webmaster Tools (available at bing.com/webmasters) is Bing's equivalent of Google Search Console. It allows you to submit your sitemap, check whether pages are indexed, and see any crawl errors. Submitting your sitemap to Bing Webmaster Tools is the most reliable way to ensure your content is available to the Bing index - and therefore to ChatGPT Search.

If your site is not indexed by Bing, ChatGPT Search cannot cite it regardless of how well-structured the content is. Bing indexing is the starting point.

What Affects Source Selection

OpenAI has not published detailed documentation on exactly how ChatGPT Search selects which pages to cite from Bing's results. Based on observable patterns, the following factors appear relevant.

Bing search ranking. Pages that rank well in Bing's traditional search results for a query are more likely to be retrieved and cited. Bing's ranking signals are broadly similar to Google's - content quality, backlinks, page authority, and technical performance all matter.

Content clarity and directness. ChatGPT Search extracts and synthesises passages from pages. Content that states its key information clearly and early is easier to extract from. A page that leads with a direct answer or definition gives ChatGPT Search more to work with than a page that builds to its main point over several paragraphs.

Freshness. ChatGPT Search is often triggered by queries about current information. Pages that are regularly updated and have recent publication or modification dates may be prioritised for time-sensitive queries. Keeping content current is particularly relevant for topics that change frequently.

Site credibility signals. Authoritative sites with clear authorship, consistent publishing, and established content depth appear to be preferred sources. A site with ten well-maintained articles on a specific topic is a more reliable source than a site with one article.

Not blocking OAI-SearchBot. OpenAI uses its own crawler, OAI-SearchBot, to index content for use in ChatGPT Search. If OAI-SearchBot is blocked in your robots.txt file, your content will not be available for ChatGPT Search citations. Check your robots.txt to confirm this crawler is not excluded.

Practical Steps for ChatGPT Search Visibility

Submit to Bing Webmaster Tools. If you have not already, create an account at bing.com/webmasters and submit your sitemap. This ensures Bing is aware of your pages and has a path to index them.

Check your robots.txt file. Confirm that neither Bingbot nor OAI-SearchBot is blocked. A robots.txt entry that blocks all unrecognised bots or that specifically excludes these crawlers will prevent ChatGPT Search from accessing your content.

Keep content current. For topics where information changes over time - AI tools, software, regulations, pricing - update articles when the information changes. Outdated content is a weaker source for a system designed to retrieve current information.

Write with direct, extractable statements. Each main section of an article should open with a direct statement of the key point for that section. ChatGPT Search extracts passages, not full articles. A passage that makes its point clearly in the first sentence is more useful than a passage that requires reading to the end to understand.

Build FAQ sections with specific answers. FAQ content matches the question-and-answer format that AI search is designed to serve. Direct, specific FAQ answers are frequently extracted for AI responses. For guidance on structuring and marking up FAQ content, read How to Add FAQ Schema to Your Pages.

Build topical depth. A cluster of well-linked articles on a specific topic signals consistent expertise. For guidance on building this structure, read How to Track Your AI Search Visibility to understand how to measure whether your depth is producing results.

How to Check If ChatGPT Is Citing Your Site

Manual testing is the most direct approach.

Search for your core topics in ChatGPT with the search feature enabled. Review the response and check the source citations listed alongside or beneath the answer. If your domain appears in the citations, your content is being used as a source for those queries.

Test multiple phrasings of your core topics. ChatGPT Search results vary by query phrasing, and a site may be cited for one phrasing but not another.

Keep a simple record of which queries return citations to your site. This provides a baseline to track whether your visibility improves over time as you add content and improve your Bing indexing.

Common Mistakes

Ignoring Bing indexing entirely. Many site owners submit to Google Search Console and stop there. Since ChatGPT Search runs on the Bing index, not Google's, Bing indexing is a separate and necessary step. Submit your sitemap to Bing Webmaster Tools if you have not done so.

Blocking OAI-SearchBot. OpenAI's crawler must be allowed access to your content for it to appear in ChatGPT Search. A blanket bot-blocking rule in robots.txt can exclude it inadvertently.

Publishing content and not updating it. For topics where information changes, ChatGPT Search will prefer current sources. An article published in 2024 with outdated information is a weaker source than a recently updated article.

Expecting the same results as Google. ChatGPT Search and Google AI Overviews draw from different indexes and use different systems. A site that appears frequently in Google AI Overviews is not guaranteed the same visibility in ChatGPT Search, and vice versa. Track each platform separately.

Frequently Asked Questions

Does ChatGPT Search use Google's index? No. ChatGPT Search draws from the Bing web index, not Google's. Being indexed by Google does not guarantee availability to ChatGPT Search. Submitting to Bing Webmaster Tools is a separate step required for ChatGPT Search visibility.

How do I know if OAI-SearchBot is blocked on my site? Open your robots.txt file (accessible at yourdomain.com/robots.txt) and check for any rules that block OAI-SearchBot by name, or blanket rules that block all bots except specifically allowed ones. If OAI-SearchBot is not in an allow list, it may be excluded.

Does ChatGPT Search always use web results? No. ChatGPT determines whether to use web search based on the query. Questions about current events, specific facts, recent news, and how-to tasks frequently trigger a web search. Abstract questions, creative tasks, and queries about general knowledge that does not change often may be answered from the model's training data without a web search.

Will optimising for ChatGPT Search also help with Perplexity? Partially. The content quality factors - clarity, directness, FAQ structure, topical depth - apply across both. The indexing factors differ: Perplexity uses its own crawler (PerplexityBot), while ChatGPT Search uses Bing and OAI-SearchBot. Confirm both crawlers are allowed in your robots.txt for maximum coverage.

Is there a way to opt out of ChatGPT Search citations? Yes. You can block OAI-SearchBot in your robots.txt file to prevent your content from being used in ChatGPT Search. This is a deliberate choice some publishers make for editorial or licensing reasons. For most small business websites, being cited is a benefit rather than a concern.

Summary

ChatGPT Search retrieves live web content from the Bing index to supplement AI responses and cites the sources it uses. Being indexed by Bing is a prerequisite for ChatGPT Search visibility.

Submit your sitemap to Bing Webmaster Tools. Check that neither Bingbot nor OAI-SearchBot is blocked in your robots.txt file.

Content that states key information directly, has clear heading structure, and includes well-written FAQ sections is easier for ChatGPT Search to extract and cite.

Keep content updated on topics where information changes regularly. Build topical depth through a cluster of well-linked articles on your subject.

Track appearances by manually searching your core topics in ChatGPT with search enabled and checking the source citations in the response.

For the broader GEO strategy, read What is Generative Engine Optimisation (GEO)?