ChatGPT Ranking: How to Get Indexed and Cited by ChatGPT (2026)

ChatGPT ranking is the order in which ChatGPT search picks and cites web pages, and OpenAI says only that it “ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed.” There is no published list of ranking factors and no URL submission tool. To get indexed by ChatGPT, OpenAI’s ChatGPT search help article asks for two things: allow OAI-SearchBot to crawl your site, and make sure your host or CDN lets in traffic from OpenAI’s published searchbot IP addresses.

Everything else on this page builds on those two rules. It covers where ChatGPT gets its results, the four OpenAI crawlers, why a page goes missing from OpenAI’s index (in OpenAI’s own words), a 12-check audit you can run this afternoon, and how to measure ChatGPT traffic in GA4. Mechanics come only from OpenAI’s documentation, read on November 1, 2026. Third-party studies are labeled with their method. It is the ChatGPT chapter of our answer engine optimization guide.

How to get indexed by ChatGPT: what OpenAI actually asks for

OpenAI has no Search Console. You cannot submit a sitemap to it, request indexing, or see a coverage report. Getting indexed by ChatGPT means being reachable by OpenAI’s search crawler and discoverable from pages it already knows. OpenAI’s documentation gives you four concrete rules:

1. Allow OAI-SearchBot in robots.txt. OpenAI’s crawler page says “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links.”
2. Allow OpenAI’s IP ranges at the firewall. The same page recommends “allowing requests from our published IP ranges.” A Cloudflare or Sucuri rule that challenges unknown bots blocks OpenAI even when robots.txt says Allow.
3. Wait about a day after a robots.txt change. OpenAI says “it can take ~24 hours from a site’s robots.txt update for our systems to adjust.”
4. Serve the content without a login or a script. OpenAI’s list of reasons a page is missing or stale (below) includes login walls, personalization and content that “depends heavily on scripts or dynamic loading.”

Does ChatGPT have its own index?

Yes. OpenAI confirms it in its help article on offline web search, a mode for some Enterprise, Edu, Healthcare and regulated workspaces. In that mode ChatGPT uses “OpenAI’s indexed and cached web content instead of using live web search at the time of each request.” The same article is the closest thing OpenAI publishes to an indexing guide. It says “there is no refresh SLA for a specific URL” and lists the common reasons content is missing or stale:

OpenAI also warns that in offline mode “deeper navigation can be limited to pages that are already indexed or cached.” That is a direct reason to keep your most important answers on pages that sit close to your homepage.

Where ChatGPT search gets its results

The common claim that “ChatGPT search is just Bing” is not what OpenAI says. Its documents describe several inputs and never give the share of each. Here is every source OpenAI names, with where it says so:

So Bing is a documented input, not a documented gate. For the consumer product OpenAI does not say a page must be in Bing to appear. Being indexed in Bing is still cheap insurance, because Microsoft is the only general web search company OpenAI names. Verify your site in Bing Webmaster Tools and push changed URLs with IndexNow, which the IndexNow FAQ says reaches Bing, Amazon, Naver, Seznam, Yandex and Yep (not Google), while noting “Submitting a URL does not guarantee immediate indexing.” Our Bing Webmaster Tools and IndexNow setup guide has the steps.

ChatGPT rewrites your buyer’s question before it searches

This is the most useful line in OpenAI’s help center for anyone trying to rank, and almost nobody quotes it. When ChatGPT uses a search partner, it “typically rewrites your query into one or more targeted queries.” OpenAI’s example: a researcher asks “what’s the latest on the development of drugs that target CCR8 for cancer?” and ChatGPT first searches “CCR8 immunotherapy drug development 2025,” then follows up with “CHS-114 conference 2025.”

Two consequences for your pages:

• You compete on keyword-style queries, not on the chat prompt. The rewritten query is short, uses the entity name and often adds a year. A page titled “Estate planning attorney fees in Ohio (2026)” matches that shape. A page titled “Our Thoughts on Planning” does not.
• Location and memory change the query. OpenAI says ChatGPT may rewrite “good restaurants near me” into “top restaurants San Francisco” from the user’s approximate IP location, and may use saved memories (“good vegan restaurants San Francisco”). If you serve a city, the city name belongs in the title, the first paragraph and the service descriptions.

The four OpenAI crawlers and what each one controls

OpenAI documents four user agents on its crawler page. One decides whether ChatGPT search can show you. One decides training. One acts for a live user. One only checks ads.

The operating rules from the same page:

• Each setting is independent. You can allow OAI-SearchBot and disallow GPTBot. If you allow both, OpenAI “may use the results from just one crawl for both use cases.”
• Each bot has its own published IP list: searchbot.json, gptbot.json, chatgpt-user.json and adsbot.json.
• When OAI-SearchBot and GPTBot fetch robots.txt itself, they may add a robots.txt marker to the user-agent string, so you can tell those hits apart in logs.
• The current example string for search is Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36; compatible; OAI-SearchBot/1.4; +https://openai.com/searchbot. OpenAI notes “the version number may change,” so match on OAI-SearchBot, not the full string.

Here is a robots.txt that keeps you rankable in ChatGPT and opts out of training. It matches the template in our AI Search Readiness Study 2026; delete the GPTBot lines if you are happy for your content to be used in training.

# ChatGPT search: allow (required to appear in ChatGPT search answers)
User-agent: OAI-SearchBot
Allow: /

# User-initiated fetches: allow (OpenAI says robots.txt may not apply anyway)
User-agent: ChatGPT-User
Allow: /

# Model training: your decision (this example opts out)
User-agent: GPTBot
Disallow: /

# Everyone else, including Googlebot and Bingbot
User-agent: *
Allow: /

Sitemap: https://www.example.com/sitemap.xml

Most small firms have never written any of this. In the study’s crawl of 419 live US small-business websites in July 2026, 97.9% of robots.txt files did not mention OAI-SearchBot, none blocked it, and 91.2% stated no policy on GPTBot. The risk for most firms is not a deliberate block. It is a firewall default nobody remembers switching on.

How to keep a page out of ChatGPT search

Disallowing OAI-SearchBot stops OpenAI from using your content in answers, but OpenAI says the page “can still appear as navigational links.” OpenAI’s publisher FAQ, as captured in a web.archive.org snapshot dated October 4, 2026, explained why: if OpenAI obtains a disallowed URL “from a third-party search provider or by crawling other pages,” it may show “just the link and page title.” Its fix was the noindex meta tag, with a catch: “in order for our crawler to read a meta tag, it must be allowed to crawl the relevant page(s).” So to hide a page completely, allow OAI-SearchBot on that URL and serve noindex. That FAQ was written for the ChatGPT Atlas browser, and the original URL now redirects to OpenAI’s Atlas retirement notice, so treat it as OpenAI’s last written guidance rather than a live page.

What is not on OpenAI’s crawler page: llms.txt. It is a proposal by Jeremy Howard published September 3, 2024 (llmstxt.org), not a ratified standard. OpenAI publishes an llms.txt for its own developer docs, yet its crawler documentation says nothing about reading yours. Our SEO terms glossary tracks its status, including Google’s statement that Google Search ignores these files.

What decides ChatGPT ranking once you are indexed

OpenAI’s whole public statement on ranking is the sentence quoted at the top: multiple factors, relevance and reliability, no guaranteed placement. Anyone who hands you a weighted list of “ChatGPT ranking factors” is guessing. What you can work with is what OpenAI documents about when ChatGPT searches, plus third-party measurement that is honest about its method.

When ChatGPT decides to search at all

OpenAI says ChatGPT “may search the web automatically when your question would benefit from current information,” and users can force it through View all tools, then Search, or by typing / and choosing Search. Web search is available on Free, Go, Plus, Pro, Business, Enterprise and Edu, and to signed-out users. ChatGPT is, in effect, classifying each prompt as “needs the web” or “answer from memory.” A prompt it answers from memory has no citations, so no page ranks for it. Prompts about current prices, rules, dates, named local firms and this year’s numbers are the ones that trigger a search.

How citations appear in ChatGPT

Per the help center, responses that use search “may include citations.” Users select a citation to open the source, can hover over one on desktop web to preview it, and can open Sources to see cited sources “and other relevant links.” Images in a response carry their source, and on iOS and Android local results may include a map. Being in the Sources panel is not the same as being an inline citation; your monthly log (below) should record which one you got.

What third-party data shows

Two studies with stated methods are worth knowing. Both measure correlation, not OpenAI’s actual ranking code.

Read together with OpenAI’s query-rewriting note, the practical rules are short. Title pages in the words of the rewritten query. Use readable slugs such as /estate-planning-attorney-fees-ohio/, not /?p=4182. Put the fact in the first sentence under each heading, with the year when it changes yearly. And accept that for professional services most ChatGPT answers carry no link at all, which is why the ones that do are worth fighting for. My view: for a 10 to 50 person firm, fixing access and titles on your 20 highest-impression pages beats any amount of “AI content” production.

The ChatGPT indexing and ranking audit: 12 checks

Run these per page, starting with pages that already get impressions in Search Console. Checks 1 to 6 are access: fail one and ChatGPT may never see the page. Checks 7 to 12 decide whether ChatGPT picks your page over another one it found. The structure mirrors our Claude citation eligibility audit, so you can run both in one pass.

# Check 2: does your firewall let an OAI-SearchBot request through?
curl -s -o /dev/null -w "%{http_code}\n" \
  -A "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36; compatible; OAI-SearchBot/1.4; +https://openai.com/searchbot" \
  https://www.example.com/your-page/

# Check 7: is the answer in the raw HTML?
curl -s https://www.example.com/your-page/ | grep -c "a sentence copied from your page"

# Check 12: has OAI-SearchBot visited in your logs?
grep "OAI-SearchBot" /var/log/nginx/access.log | awk '{print $9, $7}' | sort | uniq -c | sort -rn | head

A faked user agent only tests rules keyed on the bot’s name. Real OpenAI traffic comes from the IP ranges in the JSON files, which is why check 3 exists. If you want someone to run all twelve with you and fix the failures, you can book a consultation.

How to check whether ChatGPT has indexed a page

There is no official index checker, so use these tests in order of strength:

1. Server logs. OAI-SearchBot requests to the URL, with 200 responses from IPs in searchbot.json, prove OpenAI has crawled it. They do not prove it ranks.
2. Offline web search, if you have it. In an eligible Enterprise or Edu workspace with offline web search on, ask ChatGPT about the URL. OpenAI says it “can use that URL only if the page is available in the index or cache.” A refusal means the page is not there.
3. A live search for your exact title. Turn on Search, ask for your page’s title in quotes, and see whether your URL appears in the citations or Sources.
4. Not a valid test: pasting your URL. That triggers ChatGPT-User, which OpenAI says “is not used to determine whether content may appear in Search.” ChatGPT reading a pasted link tells you nothing about indexing.

How to measure ChatGPT traffic and citations in GA4

Start with the scale. OpenAI’s last written figure is “more than 900M weekly active users” in its February 27, 2026 post. Sam Altman cited 1.2 billion weekly users on stage at DevDay on September 29, 2026, as reported live by Engadget. Reach is not traffic, and the traffic data is noisy. StatCounter’s AI chatbot referral data puts ChatGPT at 80.47% of AI chatbot referrals in September 2026, ahead of Gemini at 10.94%; its October 2026 reading breaks sharply (ChatGPT 51.03%), so wait for the series to settle before quoting it. Our AI search statistics roundup keeps the full series with dates and methods.

The utm_source=chatgpt.com tag. OpenAI’s publisher FAQ said “ChatGPT automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs” (archived October 4, 2026). That page now redirects to the Atlas retirement notice, and I found no current OpenAI page that restates it. Check your own data: in GA4, open Reports, Acquisition, Traffic acquisition, set the dimension to Session source / medium and search for chatgpt.com. If the tag is still being added, you will see chatgpt.com / referral or a source of chatgpt.com on landing URLs with the parameter.

The AI Assistant channel. On May 13, 2026, Google Analytics added an AI Assistant channel to the Default Channel Group (GA4 release notes), and on June 11, 2026 a Source Group dimension with built-in grouping “for emerging traffic sources like ChatGPT (OpenAI) and Perplexity.” Per Google’s channel definitions, the channel matches medium “ai-assistant” or a referrer on Google’s list, with ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples. It excludes AI Overviews and AI Mode, which stay in Organic Search.

1. Go to Reports, Acquisition, Traffic acquisition, with the primary dimension set to Session default channel group, and look for AI Assistant.
2. Switch to Session source and search for chatgpt.com. If it still shows under Referral, the default list is missing it.
3. If so, go to Admin, Data display, Channel groups, Create new channel group. Add a channel named “AI search” with the condition Source matches regex .*(chatgpt\.com|openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai).*, then click Reorder and drag it above Referral. Google’s channel groups help explains why: traffic goes to “the first channel whose definition it matches.” Standard properties allow 2 custom groups of up to 50 channels, and custom groups apply to past data.
4. Add a Landing page dimension. Similarweb’s weekly data shows the share of ChatGPT referrals landing on homepages jumped from roughly 26 to 29% to about 62 to 63% by late May 2026, so a cited article may show its traffic on your homepage.

Our guide to tracking AI search traffic in GA4 and Search Console covers the reporting layer in more depth.

A monthly prompt log. Referrals miss citations nobody clicks, so check ChatGPT directly. Pick 10 to 20 questions your buyers ask, in their words, and run the same set on the first working day of each month with Search on. Copy this log:

If “Did ChatGPT search?” is No, no page can rank for that prompt, so rephrase it around something current or drop it. If ChatGPT searched and cited competitors, compare their page with yours on checks 7 to 11.

Frequently asked questions

How do I get indexed by ChatGPT?

OpenAI has no submission tool. Its help center says to allow OAI-SearchBot in robots.txt and make sure your host or CDN allows traffic from the IP addresses in openai.com/searchbot.json. Then keep the content in plain HTML, outside logins, and linked from pages OpenAI already knows. Robots.txt changes take about 24 hours to register, and OpenAI gives no refresh deadline for any single URL.

How does ChatGPT ranking work?

OpenAI says only that ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information, and that placement is not guaranteed. It publishes no factor list. It does document that ChatGPT rewrites prompts into short, targeted search queries, often adding a place or year, so pages whose titles match that query shape and state the fact up front have the best chance.

Does ChatGPT use Bing or Google for search?

OpenAI says ChatGPT search sometimes partners with other search providers and points to Microsoft's and Shopify's privacy policies. For Enterprise and Edu workspaces it names Bing directly. It also crawls with OAI-SearchBot and keeps its own indexed and cached web content. OpenAI does not name Google as a provider or say what share of results comes from each source.

How do I rank on SearchGPT?

SearchGPT was OpenAI's search prototype, announced July 25, 2024. OpenAI brought it into ChatGPT as ChatGPT search on October 31, 2024, so ranking on SearchGPT now means appearing in ChatGPT search. The steps are the same: allow OAI-SearchBot, let OpenAI's IP ranges through your firewall, and write pages whose titles and first sentences answer the query directly.

Will blocking GPTBot remove my site from ChatGPT?

No. OpenAI treats GPTBot and OAI-SearchBot as independent settings. GPTBot collects content that may be used to train its models, while OAI-SearchBot decides whether you appear in ChatGPT search answers. You can disallow GPTBot to opt out of training and still allow OAI-SearchBot to stay visible in search. Blocking OAI-SearchBot is the setting that removes you.

How can I tell if ChatGPT has indexed my page?

There is no official checker. Look for OAI-SearchBot requests with 200 responses in your server logs, verified against OpenAI's published IP list. Then run a ChatGPT search for the page's exact title and look for your URL in citations or Sources. Pasting your URL into ChatGPT proves nothing, because that fetch uses ChatGPT-User, which OpenAI says does not decide search appearance.

Does llms.txt help ChatGPT ranking?

Nothing in OpenAI's crawler or ChatGPT search documentation says it reads a site's llms.txt for search. llms.txt is a proposal from September 2024, not a ratified standard, and Google says Google Search ignores it. Adding one does no harm, but allowing OAI-SearchBot, fixing firewall rules and serving content in plain HTML are the documented steps.