TL;DR
- ChatGPT reads Bing. Seer Interactive found 87% of SearchGPT citations match Bing's top results against only 56% for Google. If you are not in Bing's index, you cannot be cited.
- OpenAI's crawlers do not execute JavaScript. Content rendered in the browser rather than served in the HTML is invisible to them.
- Only about 18% of ChatGPT conversations trigger a web search, and the opening question is roughly 2.5 times more likely to produce citations than turn ten (Profound, ~700,000 conversations, October to December 2025).
- Two thirds of the citation pool is unreachable. Ahrefs found 67% of ChatGPT's top 1,000 cited pages are Wikipedia (29.7%), brand homepages (23.8%) and app store listings (6.6%).
- Referring domains were the single strongest predictor of citations in SE Ranking's analysis of 216,000 pages. Off-page work still decides this.
This is the assistant half of a wider discipline. The research base behind it, including which content methods have controlled evidence, sits in our guide to generative engine optimization.
How do you get your website cited by ChatGPT?
Be indexable by Bing, be readable without JavaScript, and allow OpenAI's crawlers. Those are eligibility conditions rather than optimisation tactics, and failing any one of them makes the rest of your work irrelevant.
Two plain-English terms first:
- A crawler is software that visits web pages automatically. OpenAI runs several, each with a different job.
- Client-side rendering means your page arrives at the browser mostly empty and JavaScript fills in the content afterwards. Humans see a full page. A crawler that does not run JavaScript sees the empty one.
Once eligible, the work is mostly off your own site, which is the part most programmes underfund.
Why does ChatGPT cite Bing results, not Google?
Because that is the search index behind it. How to get cited by ChatGPT starts with a fact almost no guide leads with. Seer Interactive's controlled study found 87% of SearchGPT citations match Bing's top results, while only 56% match Google's.
87% against 56%: the share of ChatGPT search citations matching Bing's top results versus Google's (Seer Interactive)
That single difference explains a great deal of unexplained invisibility. A site can rank well in Google, do everything a Google-focused guide recommends, and remain absent from ChatGPT because nobody ever checked Bing.
Two things to do about it, both free:
Register with Bing Webmaster Tools at bing.com/webmasters. Verify your site, submit your sitemap, and check the Site Explorer and Index Explorer reports for pages Bing has not indexed. Most teams have never opened it.
Use IndexNow. It is a protocol Bing supports that lets you notify search engines immediately when a page changes, rather than waiting for a crawl. Bing Webmaster Tools can generate the key and many content platforms support it with a setting rather than code.
This is the clearest practical difference between ChatGPT visibility and Google visibility, and it is covered from the other direction in our guide to ranking in AI Overviews, where the index is Google's own.
What does ChatGPT actually cite?
How to get cited by ChatGPT depends on what it cites, and that is a small, concentrated set of sources, most of which you cannot become. This section is the honest deflator and it should shape your expectations before you spend anything.
Ahrefs analysed ChatGPT's top 1,000 cited pages and found 67% sit in categories most brands cannot reach:
| Category | Share of top cited pages |
|---|---|
| Wikipedia | 29.7% |
| Brand homepages | 23.8% |
| App store listings | 6.6% |
| Combined | 67% |
Look closely at the second row, because it is the reachable one. Brand homepages are 23.8%, which means ChatGPT frequently cites a company's own front page when answering about that company. Your homepage being clear, current and factually complete about what you do is doing more work than most of your blog.
The wider pattern holds across engines. Analyses synthesising hundreds of millions of citations consistently find a small group of domains taking a large majority of citation share, with Reddit, Wikipedia, YouTube and LinkedIn recurring at the top.
What this means for planning: you are not competing for the whole citation pool. You are competing for the remaining third, and within that, for the specific prompts your buyers actually type. Any vendor promising broad ChatGPT visibility is selling against a pool that is mostly closed.
The five prerequisites, before any content work
Time required: about 90 minutes to check all five. You need: access to your site's code or hosting settings, Bing Webmaster Tools, and your robots.txt file.
1. You are in Bing's index. Search site:yourdomain.com on Bing. If few or no pages appear, nothing else in this article will help. Fix this first.
2. Your pages work without JavaScript. Disable JavaScript in your browser and load a page you want cited. If the content disappears, OpenAI's crawlers see the same emptiness. Server-side rendering or static generation fixes it.
3. You are not blocking OpenAI's crawlers. Check robots.txt and any CDN or firewall rules for these user agents:
| Bot | What it does |
|---|---|
| OAI-SearchBot | Surfaces sites in ChatGPT search. Blocking this removes you from citations |
| GPTBot | Crawls for model training. Blocking it does not affect search citations |
| ChatGPT-User | Fetches a page when a user's prompt requires it in real time |
The distinction matters commercially. Teams that decided to block AI training frequently blocked all three, removing themselves from ChatGPT citations as a side effect they did not intend. If you want the citations but not the training use, allow OAI-SearchBot and ChatGPT-User, and decide about GPTBot separately.
4. Your pages return a 200 and are not noindex. Basic, and worth confirming on the specific pages you care about rather than assuming.
5. Your company name and description are consistent everywhere. Site, LinkedIn, directories and any press. Inconsistency makes it harder for a model to resolve that all those mentions describe one company.
How to get cited by ChatGPT: the seven steps
Do these in order. The first three are eligibility, the last four are the actual competition.
Step 1. Get into Bing's index
Verify your site in Bing Webmaster Tools, submit your XML sitemap, then use URL Inspection to check specific pages. Where pages are missing, use URL Submission to request indexing directly.
Step 2. Serve your content in the HTML
If your site renders client-side, move the pages you want cited to server-side rendering or static generation. This is a development task rather than a marketing one, and it is the single most common invisible blocker.
Step 3. Allow the right crawlers
Confirm OAI-SearchBot and ChatGPT-User are permitted in robots.txt and at your CDN. Check the CDN separately, because bot-protection rules frequently block crawlers that robots.txt allows.
Step 4. Make your homepage do its job
Given that brand homepages are 23.8% of ChatGPT's top cited pages, treat yours as a reference document rather than a brochure. State plainly what the company does, who it serves, where it operates and when it was founded. Vague positioning language is not citable.
Step 5. Earn referring domains
SE Ranking's analysis of 216,000 pages found referring domains were the single strongest predictor of ChatGPT citations. These are other websites linking to you. Digital PR, original research, industry directories and genuine community participation all build them. This is slow and it is the highest-value work in this list.
Step 6. Be present where ChatGPT already looks
Identify the domains cited in answers to your buyers' questions, then work on your presence there rather than only on your own site. Review sites, comparison pages, directories and communities. Sources travel in packs, so appearing alongside the domains already trusted in your category matters more than outranking them.
Step 7. Write passages that survive being quoted alone
Assume any 200 to 300 word block is retrieved without its surroundings. Name the subject in the first sentence, put the number and the source inside the block, and answer before elaborating. The content methods with controlled evidence behind them are covered in the generative engine optimization guide.
Which pages get cited by ChatGPT, and when?
Mostly on the opening question of a conversation. Getting cited by ChatGPT is a turn-one problem, and only in the minority of conversations that search at all.
Profound analysed roughly 700,000 ChatGPT conversations from October to December 2025 and found:
- About 18% of conversations trigger at least one web search. The rest are answered from the model's own knowledge, with no citation opportunity.
- The first turn is roughly 2.5 times more likely to produce citations than turn ten, and close to four times more likely than turn twenty.
- Citations arrive in packs. ChatGPT cites several sources side by side rather than picking one winner, which means competitors appear together and your goal is inclusion rather than exclusivity.
The planning consequence: target the questions people open with. "Best CRM for a small sales team" is an opening question. "How do I export from that one" is a follow-up and rarely triggers a search. Opening questions are broader, more commercial and more competitive, which is exactly where the effort belongs.
How do you measure ChatGPT citations?
There is no Search Console for ChatGPT, so measuring how to get cited by ChatGPT is something you build. You build the measurement yourself.
Run a fixed prompt set. Take 20 to 50 questions your buyers actually ask, run them monthly in ChatGPT, and record which brands and domains appear. Keep the wording identical between runs so the comparison means something.
Vary the reasoning setting. Source sets differ substantially between modes, so a single configuration gives you a partial picture. Record which setting you used.
Calculate share of mentions. Prompts where your brand appears, divided by total prompts. Track the direction monthly rather than reading any single month as a result.
Check GA4 referrals by filtering for chatgpt.com. Treat this as a floor rather than a measure, because a large share of assistant traffic arrives without a referrer and lands as direct.
One check is not a measurement. ChatGPT is non-deterministic, so the same prompt can return different sources on consecutive runs. Three monthly readings are a trend; one reading is an anecdote.
What do most teams get wrong?
Optimising for Google and assuming ChatGPT follows. The 87% Bing match against 56% for Google is the whole reason this fails.
Blocking all OpenAI crawlers to prevent training. A common and self-defeating decision. Blocking GPTBot affects training. Blocking OAI-SearchBot removes you from citations. They are separate choices.
Client-side rendering. Your page looks perfect to you and is empty to the crawler. Test with JavaScript disabled rather than assuming.
Expecting to displace Wikipedia. Roughly 30% of ChatGPT's top cited pages are Wikipedia. Be the source cited next to it, not instead of it.
Neglecting the homepage. It is 23.8% of the top cited pages and is usually the least factual page on the site.
Checking once and forming a belief. Non-determinism means one run tells you nothing.
Treating this as a content problem. Three of the five prerequisites are technical, they are free to fix, and no amount of writing compensates for failing them.
How does Euracle approach ChatGPT visibility?
Euracle treats how to get cited by ChatGPT as an engineering question first: the first deliverable is a technical eligibility report, not a content plan.
The Eureka Method, Euracle's discovery sprint, runs four phases here.
Discover checks the five prerequisites before anything else and reports whether the site is eligible to be cited at all, which frequently answers the client's question on its own.
Design separates the free technical fixes from the slow off-page work and sequences them accordingly.
Deploy ships the eligibility fixes first, then the homepage rewrite, then the off-page programme.
Scale runs the prompt set monthly at more than one reasoning setting and reports share of mentions against competitors.
The stack is ordinary: Bing Webmaster Tools and Google Search Console for index coverage, GA4 for referral floors, and n8n with the Claude API to run the monthly prompt set on a schedule rather than by hand.
Two structural commitments come from how Euracle is set up. Senior practitioners only: the people in the pitch do the work, which matters here because three of the five prerequisites are engineering questions rather than marketing ones. And one contract across six disciplines, so a finding that the real blocker is client-side rendering does not require handing the client to a different vendor to fix.
If you want the eligibility check run before anyone quotes you a content programme, that is Euracle's AEO and GEO service. The rendering problem is most common at B2B SaaS companies, where marketing sites are frequently built as single-page applications.
If you are still deciding which term describes this work, our guide to AEO vs GEO vs SEO settles the vocabulary, and answer engine optimization covers how the Google and assistant halves should split a budget.
FAQ
Conclusion
You can now check how to get cited by ChatGPT in about ninety minutes whether ChatGPT is able to cite you at all. Search Bing for your domain, load a page with JavaScript disabled, and read your robots.txt and CDN rules for OAI-SearchBot. Those three answers explain most unexplained invisibility, and all three are free to fix. Then treat your homepage as a reference document, target the questions people open conversations with, and put the remaining effort into referring domains and third-party presence, because that is what the measurement points at. If you want the eligibility check run before anyone quotes you a content programme, talk to Euracle about AEO and GEO.
Sources
- Seer Interactive, controlled study of SearchGPT citations. 87% matching Bing's top results against 56% for Google.
- Ahrefs, analysis of ChatGPT's top 1,000 cited pages. Wikipedia 29.7%, brand homepages 23.8%, app store listings 6.6%, combining to 67%.
- Profound, How ChatGPT sources the web, February 2026. Approximately 700,000 US English-language ChatGPT conversations, October to December 2025. The 18% search rate, the turn-one citation multiplier and the citation-pack finding. https://www.tryprofound.com/blog/chatgpt-citation-sources
- SE Ranking, analysis of 216,000 pages finding referring domains the strongest predictor of ChatGPT citations.
- OpenAI crawler documentation, naming OAI-SearchBot, GPTBot and ChatGPT-User and their distinct purposes.
