Query. how do you get cited by chatgpt

Guides26 min read

How do you get cited by ChatGPT?

A results list with two rows marked blue, joined by dotted lines to two highlighted lines in an answer card with amber citation tags.

Getting cited by ChatGPT means OpenAI’s ChatGPT links your page as a source in its answer. In ChatGPT search answers, that happens after its web search finds the page for one of the searches it runs. The link is called a , and it lets a reader open the page behind the answer.

A citation passes two gates in order. Retrieval comes first, when ChatGPT’s search reaches the page and returns it for one of its searches. Selection comes second, when ChatGPT picks a few sources from what came back. Content that OpenAI’s partners supply directly and product data from merchants reach answers by other routes.

Most figures here come from studies published between March and November 2026, and many describe free, logged-out sessions. Much of the data predates GPT-6. OpenAI began rolling out GPT-6 on October 7, 2026, according to SISTRIX, so the figures describe ChatGPT as those studies found it.

Readers new to AI search can start with What is AI? and the Answer Engine Optimization (AEO) guide.

ChatGPT cites pages its search step finds

ChatGPT search cites a page when its web search for a question returns that page and ChatGPT uses it as a source. The search is the retrieval gate, and the choice of source is the selection gate.

ChatGPT decides for each question whether to search at all, according to OpenAI’s ChatGPT search help article. The same article says that when ChatGPT searches, it ranks the results on many factors meant to bring up reliable, relevant information.

Ahrefs’ ChatGPT fan-out study looked at 66.6 million free, logged-out ChatGPT responses from August and September 2026. Of those responses, 14.4% contained fan-out queries, the background searches explained under ChatGPT turns one question into one or more searches. The study doesn’t show how many of the other 85.6% searched.

An answer written without a web search has no web sources to link.

Two gates to a citation

How one question passes the retrieval and selection gates

  1. Question

    A person types a question. ChatGPT first works out whether answering it calls for a web search.

  2. Searches

    To search, it writes one or more short queries of its own. A page has to match those queries, and they can use words the person never typed.

  3. Retrieval

    Each query brings back a list of results. Only pages on those lists move on to the next gate.

  4. Selection

    ChatGPT judges the results mostly by their titles and snippets and picks a few. The rest of the list goes unused.

  5. Citation

    The picked pages show up as links beside the answer and in its Sources panel, where a reader can open them.

ChatGPT chooses its sources from the pages retrieval brings back, and the answer links the ones it picks.

What is a ChatGPT citation?

A ChatGPT citation is a link in an answer to a page the answer drew on, shown beside the text and in its Sources panel. The Sources panel is the side panel ChatGPT opens from an answer, and it holds the cited pages along with other related links.

Most links in the Sources panel are related links that the answer doesn’t cite. RESONEO’s ChatGPT retrieval study counted 61,332 URLs in Sources panels in July 2026. Of those, 53,716 sat in a “More” list at the bottom and were never promoted or cited.

A citation can appear with no mention of the site in the answer text. In Ahrefs’ citation stability study, 54% of the websites ChatGPT cited weren’t named in the answer text. How do you get mentioned in ChatGPT? covers names that appear without a link.

Does ChatGPT cite pages from its training data?

ChatGPT’s citations come from pages its web search returns when the question is asked. OpenAI ties that search to OAI-SearchBot, its search , and ties training to GPTBot, its training crawler.

OpenAI’s crawler page lists GPTBot for training and OAI-SearchBot for search. Each one has its own setting in . Blocking GPTBot signals that a site’s content “should not be used in training generative AI foundation models,” and it has no effect on OAI-SearchBot’s access. OpenAI’s crawler page also says that when a site allows both bots, OpenAI may use one crawl for both purposes.

Does a citation in an AI answer mean the model was trained on that page? explains how training and retrieval differ.

What is ChatGPT SEO?

ChatGPT is the work of getting a site found by ChatGPT search and cited in its answers. Most of it is ordinary search engine optimization. Like Google, ChatGPT needs a page its crawler can fetch and text that answers the question.

Three parts of ChatGPT SEO are specific to ChatGPT.

  • ChatGPT search uses OpenAI’s own crawler, OAI-SearchBot.
  • ChatGPT writes its own searches from the person’s question.
  • ChatGPT’s answers have no fixed positions a page can hold.

ChatGPT turns one question into one or more searches

ChatGPT search rewrites a question into one or more shorter searches, called , and sends them to its search sources. This step opens the retrieval gate, because a page can come back only for a search ChatGPT runs.

OpenAI’s help article says ChatGPT typically rewrites a question into one or more targeted searches for its search providers. A search provider is an outside search engine that answers ChatGPT’s searches. What is query fan-out? covers the idea across AI search systems.

In the Ahrefs fan-out study, responses that contained fan-out queries carried about 1.6 each. Ahrefs notes that paid and reasoning modes search more than the free, logged-out sessions it studied. Reasoning modes are the settings where ChatGPT spends longer working through a question before it answers.

AirOps’ retrieval and citation report, published in March 2026, found that 95% of fan-out queries had zero monthly search volume. Monthly search volume is the number of times people type a search each month, as keyword tools estimate it.

AirOps also found that 32.9% of the cited pages that appeared in Google’s top 20 came only through a fan-out query. A search for the person’s own question never brought those pages back. The report doesn’t state which ChatGPT model it tested or when it collected the data.

Keyword tools can’t list most of the searches a page competes in, because ChatGPT writes those searches as it answers.

What do ChatGPT’s fan-out queries look like?

ChatGPT’s fan-out queries are short searches, and some contain a year, the word “official”, the word “price” or a site name. In the Ahrefs fan-out study, the first three patterns each grew from August to September 2026.

Fan-out queries that contained August 2026 September 2026 Illustrative example
The year “2026” 17.6% 22.9% best project management app 2026
The word “official” 10.1% 13.1% passport renewal fee official
The word “price” 6.0% 8.9% standing desk price

A site: search is limited to one website, and 6.8% of fan-out queries used one. Of the most common site: searches, 75.9% targeted brand sites.

The study covered one model, GPT-5.6 Luna, using broad . Ahrefs also notes that OpenAI changes retrieval often, so these shares can move from month to month.

Can you see the searches ChatGPT runs?

Site owners can’t see ChatGPT’s fan-out queries. Business Hub, OpenAI’s dashboard for businesses, launched in beta on November 10, 2026, according to OpenAI’s Businesses in ChatGPT announcement. As of November 2026, it doesn’t show even the prompts behind a mention. What does Business Hub leave out? lists the rest of what the beta doesn’t report.

ChatGPT cites only pages OAI-SearchBot can reach

ChatGPT search can cite a page only when OAI-SearchBot, OpenAI’s search crawler, is allowed to fetch it and the page has no tag. A site that blocks OAI-SearchBot can appear as a plain .

Access is the first test at the retrieval gate, and it comes before any question about content. OpenAI’s crawler page says sites opted out of OAI-SearchBot “will not be shown in ChatGPT search answers.” It also says robots.txt changes take about 24 hours to reach OpenAI’s search systems.

OpenAI’s Publishers and developers FAQ says a site must not block OAI-SearchBot if it wants its content in ChatGPT’s summaries and snippets. OpenAI’s FAQ names a noindex tag as the way to keep a page out of ChatGPT search. OAI-SearchBot has to be able to fetch the page to read that tag, so a robots.txt block hides the tag from the crawler.

Robots.txt is one place that can stop OAI-SearchBot. A firewall, a security filter in front of the server, can block the crawler while robots.txt allows it. A CDN, a network of servers that delivers a site’s pages, can also block it.

OpenAI publishes the IP addresses its crawler sends requests from, the numeric addresses of its servers, at openai.com/searchbot.json. OpenAI’s crawler page recommends allowing requests from those published IP ranges as well as allowing OAI-SearchBot in robots.txt. The hosting company or a developer usually changes those rules. Your CDN can stop AI bots that ignore robots.txt covers how those rules work.

Content that appears only after a page’s scripts run may be missing from what a crawler reads. OpenAI’s crawler page doesn’t say how its crawler handles scripts. Keeping the main facts in the page’s HTML, the code the server sends before any scripts run, is a precaution. Can AI agents access JavaScript websites? covers how scripts affect AI systems.

OpenAI runs a third bot, ChatGPT-User, the fetcher that opens a page when ChatGPT reads it for a person. OpenAI’s crawler page says “robots.txt rules may not apply” to ChatGPT-User, because a user starts the request. What are GPTBot, OAI-SearchBot and ChatGPT-User? compares all three.

How do you let ChatGPT find your website?

A website lets ChatGPT find it by allowing OAI-SearchBot in robots.txt and keeping noindex off the pages it wants cited. The site also has to let OpenAI’s crawler addresses through its firewall or CDN.

  1. Open yourdomain.com/robots.txt and confirm that no rule blocks OAI-SearchBot, including the rules for all bots under User-agent: *.
  2. Open each page you want cited, right-click and choose View Page Source. Search the code for “noindex” and confirm it isn’t there.
  3. Check the firewall or CDN bot settings, or ask the hosting company, to confirm that requests from OpenAI’s published addresses get through.
  4. After any robots.txt change, allow for the delay OpenAI’s crawler page gives before you check ChatGPT again.

Check which AI bots your site allows today covers reading the file in more detail.

Server logs, the record a web server keeps of each request, show when OAI-SearchBot fetched a page and so confirm access. Any program can send OAI-SearchBot’s name. Anyone can fake an AI bot, so verify it covers checking the address behind a request.

Does ChatGPT use Bing or Google search results?

ChatGPT search uses Bing, Microsoft’s search engine, as one of its outside search providers, according to OpenAI’s help article. RESONEO’s July 2026 data also shows ChatGPT drawing on OpenAI’s own and on engines that served scraped Google results. OpenAI’s ChatGPT search launch post, from October 31, 2024, says ChatGPT search uses outside search providers along with content its partners supply directly. OpenAI doesn’t publish how it combines those sources.

About 1 in 3 of the scraped Google URLs appeared on Google’s first page, and 1 in 20 on Bing’s. RESONEO reports that free answers drew mostly on OpenAI’s own index and paid thinking-mode answers mostly on Google results. OpenAI hasn’t confirmed the split.

Two older studies measured the same overlap with Google and Bing, counting different sets of results at different times. Organic results are the regular unpaid results in a search engine’s list.

Study Data from What it counted Finding
Ahrefs’ AI search overlap study Early July 2025 ChatGPT’s in-text citations 8.00% in Google’s top 10, and 8.10% in Bing’s top 10
Seer Interactive’s SearchGPT and Bing study, historical Published February 2025, collection date not stated Citations for about 100 questions in ChatGPT search, named SearchGPT in Seer’s study 87% matched Bing’s top 20 organic results, and 56% matched Google’s

Getting a page indexed in Google and Bing opens the outside routes these studies traced. OpenAI’s own index is a separate route, and OAI-SearchBot fills it.

How does ChatGPT find new pages?

ChatGPT finds new pages through OAI-SearchBot’s crawling and its outside search providers. A pilot with Cloudflare, a company whose CDN and security services sit in front of websites, also fed network signals into OpenAI’s search index.

Cloudflare’s post on the OpenAI discovery pilot, from October 15, 2026, says the pilot ran on about 1% of Cloudflare’s network. The pilot “brought previously unknown pages into OpenAI’s search index,” according to the post. OpenAI “independently evaluates and ranks” what it finds, and the post says “a signal about a page doesn’t grant permission to access it.”

Most pages ChatGPT retrieves never get cited

Most pages ChatGPT’s search returns never reach the answer. In AirOps’ March 2026 study of 548,534 retrieved pages, 15% were cited. This is the selection gate, where ChatGPT picks a few sources from everything its searches brought back.

RESONEO counted the same gate at a finer level across 1,249 answers on free, paid and logged-out accounts. AirOps counted any citation, and RESONEO separated pages promoted to sources from pages cited in the answer text.

Stage URLs Share of retrieved URLs
Retrieved by ChatGPT’s search About 58,000 All
Promoted to a source in the Sources panel About 7,600 About 13%
Cited in the answer text About 5,000 About 9%
Opened and read About 760 About 1%

ChatGPT cites most pages from the title and snippet alone, without opening them. Of the 759 page opens RESONEO recorded, 757 came from paid accounts in thinking mode. RESONEO built these counts by reverse engineering the data behind ChatGPT’s answers, and it calls the results a snapshot.

RESONEO's count

How many retrieved URLs reach each stage

About 1 in 100 were opened and read by ChatGPT.

Each square is about 1 in 100 of the URLs ChatGPT's searches retrieved in RESONEO's sample.

Studies of the selection gate mostly find correlations. A correlation means two things rise together, and by itself it doesn’t show that one causes the other. Most of this measurement comes from companies that sell SEO tools and services.

What does ChatGPT see of a page before it cites it?

Before it picks sources, ChatGPT usually sees a page’s title and a short snippet, according to RESONEO’s study of ChatGPT’s response stream. A snippet is the short excerpt a search result shows.

For pages in OpenAI’s own index, snippets run to about 200 characters and titles pass in full. Results from scraped Google pages arrive with titles cut at 60 characters. In RESONEO’s sample, about 8 in 10 snippets from pages with an H1, the main heading on the page, contained that heading.

Snippets are frozen at crawl time. In RESONEO’s sample, about 13% ran more than a month behind the live page, so a recent edit may be missing from what ChatGPT sees.

A title tag is the page title set in the page’s code, and search results display it. In AirOps’ study, pages whose title shared half or more of its words with the search query were cited 20.1% of the time. Pages with under 10% overlap were cited 9.3% of the time.

A page’s title, its H1 and its opening lines make up most of what ChatGPT has to judge before it picks. Why does passage-level structure matter? covers writing passages that hold up when quoted alone.

Does ranking on Google help you get cited by ChatGPT?

Ranking on Google goes with more ChatGPT citations for pages ChatGPT has already retrieved. Most of what ChatGPT’s search returns comes from outside Google’s top 10.

In AirOps’ March 2026 study, which began with pages ChatGPT had already retrieved, 43.2% of pages that ranked first on Google were cited. That rate is about 3.5 times the rate for retrieved pages outside Google’s top 20.

Ahrefs’ study of ChatGPT and Google results, from August 2025, began with everything ChatGPT’s search returned. Of those results, 6.82% appeared in Google’s top 10 for the same search.

Google rankings also go with longer-lasting citations. In the Ahrefs citation stability study, 11% of pages that ranked on Google stayed cited for 7 days or more. For pages that didn’t rank, the figure was 4%.

All three findings are correlations, and none shows that a Google ranking causes a ChatGPT citation.

How do you rank in ChatGPT?

You rank in ChatGPT by getting a page returned for one of its fan-out queries and then picked as a source. Seven checks cover both gates to a ChatGPT citation lists what affects each step.

OpenAI’s help article says no site can be guaranteed top placement in ChatGPT search. ChatGPT picks its sources each time it answers, so a page has no stable position to track the way a Google ranking is tracked.

ChatGPT often cites the official source

ChatGPT’s answers to factual questions usually include an official source, such as a brand’s own site or a government page. That choice happens at the selection gate.

The Ahrefs fan-out study counted official sources two ways, by answer and by citation.

Prompt type Answers that cited at least one official source Share of citations that went to official sites
Factual 84.4% 63.4%
Comparison 68.3% 46.4%

Local questions show the same pattern in Searchable’s study of questions about 100 UK towns. ChatGPT used council or government sources for 81.3% of those questions and local news for 0.9%. Press Gazette’s report on Searchable’s local news study covers the full results across five AI tools.

In Ahrefs’ stability study, when a question named a brand, the brand’s own site stayed cited on all 20 days in 28% of cases.

A business’s own facts, such as prices, specifications, policies and dates, belong on its own site in plain text. That site is the official source for those facts. The site: searches under What do ChatGPT’s fan-out queries look like? show ChatGPT going to brand sites directly.

Do other sites’ mentions help you get cited by ChatGPT?

Brands that other sites mention frequently tend to be more visible in AI answers. Of the three platforms Ahrefs measured, ChatGPT was the least tied to those mentions.

Ahrefs’ mention types study, published November 3, 2026, covered 10,000 US brands across ChatGPT and two of Google’s AI search features. In that study, 46% of the brands had no ChatGPT mentions at all. Brand size explains part of the link, because bigger brands tend to get more of both. The authors say they “can’t say mentions cause AI visibility.”

Mentions on other sites work as corroboration, where outside sources confirm what a brand says about itself. Cited pages tend to be mentioned on other sites covers that work across platforms.

The sites ChatGPT draws on also shift when OpenAI changes retrieval. In the Ahrefs fan-out study, Reddit’s citations in US ChatGPT answers fell 88% from March to September 2026, and Wikipedia’s fell 80%.

How do you get mentioned in ChatGPT?

A business gets mentioned in ChatGPT when ChatGPT names it in an answer, with or without a link to the business’s site. Do other sites’ mentions help you get cited by ChatGPT? covers how mentions on other sites relate to a brand’s visibility in ChatGPT.

Names and links can come apart, and in Ahrefs’ stability study, 13% of ChatGPT cases lost the citation and kept the name. Another 9% kept the citation and lost the name. Business Hub counts a mention by its own rule, and What counts as a mention in ChatGPT? explains it.

To check which businesses ChatGPT names in your category, follow three steps.

  1. Ask ChatGPT who the best providers in the category are.
  2. Look up whether each business it names has third-party reviews or press coverage.
  3. Repeat the question on a few different days, given the daily changes covered under ChatGPT citations change from one day to the next.

ChatGPT citations change from one day to the next

ChatGPT’s citations change daily, and in Ahrefs’ 20-day study, 59% of the sites ChatGPT cited on day one were gone on day two. The change runs through both gates, because ChatGPT’s searches vary and so does its pick from the results.

The Ahrefs citation stability study asked five AI tools the same questions every day for 20 days. ChatGPT’s share covered 3,173 questions and 11,225 day-one citations.

Ahrefs' 20-day count

How many day-one sites ChatGPT kept citing

4 in 100 lasted all 20 days.

Each square is 1 in 100 of the sites ChatGPT cited on day one.

Measure Figure
ChatGPT-cited sites that lasted all 20 days 4.0%
ChatGPT-cited sites that lasted a week or more 13%
Dropped citations that came back within 20 days, across all five tools About 75%
Days ChatGPT ran exactly the same fan-out searches as the previous day 26%
ChatGPT-cited sites that changed when the searches stayed the same 81%
ChatGPT-cited sites that changed when the searches changed 91%
Sources in a typical ChatGPT answer 4

The 81% figure suggests part of the change happens at the selection gate. The same searches can also return different results from the search providers.

How do you measure AEO? sets out a method for logging repeated runs.

How long does it take to get cited by ChatGPT?

The time to a first ChatGPT citation varies, and the delays documented by OpenAI and the studies come from access and indexing.

A robots.txt change reaches OpenAI’s search systems after the delay OpenAI’s crawler page gives, covered under ChatGPT cites only pages OAI-SearchBot can reach. After that, the page’s snippet in OpenAI’s index can trail the live page. What does ChatGPT see of a page before it cites it? covers that lag.

Some ChatGPT citation advice has little evidence

Several popular tips for getting cited by ChatGPT rest on studies that measured something narrower than the tip claims. Some of those studies looked only at pages that were already cited. A trait those pages share may be as common on pages that were never cited.

Advice What the study measured Gate tested What it supports
Adding statistics and quotes A study by researchers at Princeton and IIT Delhi, first posted in November 2023. It rewrote a source that a simulated GPT-3.5 engine had already retrieved from Google’s top 5. Selection only Statistics, quotes and cited sources raised that page’s share of the answer by up to 40%. Keyword stuffing, repeating a search phrase over and over, gave little or no gain. The 2023 setup says little about current ChatGPT.
Updating dates Ahrefs’ freshness study from July 2025, comparing the age of cited pages Both, from cited pages only ChatGPT’s citations averaged 958 days old, against 1,432 days for Google’s organic results in the article’s text. The article cites Google’s John Mueller warning against updating publish dates without changing the content.
Posting on Reddit The Ahrefs mention types and fan-out studies Both, from visibility in answers Reddit mentions went with AI visibility across platforms, and ChatGPT’s own Reddit citations fell sharply in 2026.
Adding an llms.txt file, a proposed file that lists a site’s main pages for AI systems SE Ranking’s llms.txt study of about 300,000 domains, testing whether having the file helped a model predict AI citations Both No correlation between having the file and AI citations, and the model predicted better without the file as a factor.

Which common AEO claims hold up? rates the wider AEO claims. What is llms.txt and do I need it? covers the file in the last row.

ChatGPT shopping answers use product data

ChatGPT picks products for shopping answers from structured product data supplied by merchants and third-party providers. Structured product data means listings with fields such as price and availability. Product data gives shopping answers a retrieval route of their own, alongside the web pages OAI-SearchBot crawls.

OpenAI’s shopping help article says ChatGPT chooses product results on its own, and that they aren’t ads or shaped by OpenAI partnerships. Merchant order depends on factors such as availability, price, quality and whether the merchant makes the item or is its main seller. OpenAI says the order may change.

Merchants can also apply to send OpenAI a product feed, a file that lists each product with details such as price and availability. How do you connect a product feed to ChatGPT? covers the feed setup in Business Hub.

Does a ChatGPT business profile get you cited?

Claiming a ChatGPT business profile doesn’t change a business’s visibility in ChatGPT, according to OpenAI’s help article on setting up a business. The same article says ChatGPT may also draw on other sources about the business.

A profile lets the owner manage facts ChatGPT can share about the business, such as its website, description, category and delivery or return policies. What Businesses in ChatGPT changes for SEO covers what a profile changes for search work.

Seven checks cover both gates to a ChatGPT citation

Seven checks cover the two gates to a ChatGPT citation. The first four decide whether ChatGPT’s search can return a page, and the last three affect whether ChatGPT picks it. Passing all seven makes no promise of a citation, since OpenAI guarantees no placement.

  1. Let OAI-SearchBot in (retrieval). Confirm that robots.txt allows OAI-SearchBot, that the pages carry no noindex tag and that firewall or CDN rules let OpenAI’s addresses through. The steps are under How do you let ChatGPT find your website?.
  2. Keep the main facts in the page’s HTML (retrieval). Prices, names, dates and answers belong in the HTML the server sends.
  3. Confirm that Google and Bing have indexed the page (retrieval). OpenAI’s help article names Bing as one of its outside search providers, and the studies above trace ChatGPT results to both Google and Bing. The URL Inspection tool in Google shows whether Google has indexed a page. Bing Webmaster Tools offers a URL check for Bing.
  4. Write the title, H1 and first sentence in the words people search with (retrieval). Use the words a customer would type and the ones fan-out queries add, such as “official” or “price”, where they describe the page. Keep the main words within a title’s first 60 characters, where Google-sourced titles were cut in RESONEO’s data. Put a year in the title only when the content covers that year, because the freshness evidence doesn’t support changing dates alone. When a competitor’s page is cited, compare its title, H1 and first sentence with yours.
  5. Publish your own facts on your own site (selection). Prices, specifications, policies and dates belong on the brand’s site, where ChatGPT’s answers to factual questions tend to find an official source.
  6. Earn mentions on sites that cover your category (selection). Reviews, press and community threads confirm what the site says, and Cited pages tend to be mentioned on other sites covers that work.
  7. Check citations over repeated runs (both). Ask the same questions on several days and log the results, as Track ChatGPT citations with repeated checks describes.

Track ChatGPT citations with repeated checks

Tracking ChatGPT citations means asking the same questions in fresh chats over several days and logging which sites appear. A single answer is one sample, so each question needs several runs.

A few settings affect what a logged answer shows.

  • Sample both a logged-out session and a logged-in account, and label which mode each logged answer came from.
  • In a logged-in account, a temporary chat started without personalization doesn’t use memory or custom instructions, according to OpenAI’s Temporary Chat FAQ.
  • ChatGPT uses approximate location for local questions, according to OpenAI’s help article, so a local business should check from the area it serves.
  • RESONEO reports that tools querying OpenAI’s API, the programming interface developers use to reach OpenAI’s models, can see different results from chatgpt.com.

ChatGPT adds a utm_source tag to the links it sends. That tag is the part of a link’s address that names where a visit came from. It reads utm_source=chatgpt.com, according to OpenAI’s Publishers and developers FAQ. A referral is a visit that arrives through a link on another site, and analytics can group ChatGPT referrals by the tag.

Direct traffic covers the visits analytics can’t trace to a source. Visits from links that lose the tag can show up there, and that happened for about a week in October 2026. Search Engine Journal’s report on the ChatGPT traffic drop tied a fall in ChatGPT sessions to links that had lost the tag. SISTRIX’s review of ChatGPT in October found that most links lost it that week.

A sudden drop in ChatGPT referrals can be a tracking change, so check direct traffic for the same days before reading it as lost visibility.

For a business with a claimed profile, Business Hub Insights reports mentions and clicks from OpenAI’s own counts.

A site owner can also look in server logs for ChatGPT-User requests. In RESONEO’s data, ChatGPT opened pages almost entirely in paid thinking mode, so a ChatGPT-User request sits closer to an answer than an OAI-SearchBot crawl. Neither request shows that ChatGPT cited the page. RESONEO also reports that pages ChatGPT reads stay cached for weeks, so a repeat read may leave no new request.

How can you tell if ChatGPT cites your site?

You can tell whether ChatGPT cites your site by reading its answers to your key questions and checking your analytics.

  1. Ask ChatGPT your key questions in fresh chats.
  2. Open the Sources panel on each answer.
  3. Read the answer text as well as the Sources panel, because names and links can appear apart.
  4. Look in analytics for visits tagged utm_source=chatgpt.com, since those visits show the citations people clicked.

Frequently asked questions

Can you pay to get cited by ChatGPT?

You can’t buy a ChatGPT citation through ChatGPT ads, because OpenAI says ads don’t influence ChatGPT’s answers. In the test OpenAI announced on January 16, 2026, ads appear below the answer with a label, according to OpenAI’s ads test page.

To keep your pages out of ChatGPT search, leave OAI-SearchBot allowed and add noindex to each page. ChatGPT cites only pages OAI-SearchBot can reach explains why the crawler needs access to read the tag.

Can ChatGPT cite a page that needs a login?

OAI-SearchBot, the crawler ChatGPT search uses, can’t fetch pages behind a login. Content a partner supplies directly can reach ChatGPT search, according to OpenAI’s launch post. A person can also share a page’s text in their own chat outside ChatGPT search. Can a system reach the page? covers login walls and other blocks.