Quick Answer: Getting indexed by ChatGPT happens on your own website and in Bing Webmaster Tools, because OpenAI publishes no submission form. OpenAI's only stated requirement is that OAI-SearchBot can crawl the site and its published IP ranges are not blocked. Since OpenAI says ChatGPT search uses third-party search providers, a verified Bing property with IndexNow is the practical second lever.
"Indexed by ChatGPT" is a phrase OpenAI never uses. There is no index you submit to, no coverage report, and no inspection tool. What exists is a crawler, a set of search providers, and two documented conditions for appearing in search answers.
That is good news for a small team. The whole job is five configuration and content steps, none of which need an OpenAI account, and the only ongoing work is checking which pages get fetched and which get cited.
What does indexed by ChatGPT actually mean
Indexed by ChatGPT means your page is eligible to be retrieved and cited when ChatGPT searches the web. OpenAI's help centre states the rule in two clauses: allow OAI-SearchBot to crawl the site, and confirm your host or CDN allows traffic from OpenAI's published searchbot IP addresses.
Three OpenAI user agents touch your site, and only one of them governs search inclusion. Confusing them is the most common reason a site "does everything" and still never appears.
At a glance
| User agent | Job | Effect on ChatGPT search answers |
|---|---|---|
| OAI-SearchBot | Crawls to surface sites in ChatGPT search | Blocking it removes you from search answers (navigational links may remain) |
| ChatGPT-User | Fetches a page when a user asks about it or a Custom GPT calls it | None; OpenAI says it is not used to decide what appears in Search |
| GPTBot | Crawls content that may train foundation models | None; independent robots.txt setting |
The two conditions, verbatim
OAI-SearchBot "is used to surface websites in search results in ChatGPT's search features", per the OpenAI crawler overview. Opted-out sites "will not be shown in ChatGPT search answers", and a robots.txt change takes about 24 hours to reach OpenAI's search systems.
Beyond crawl access and IP access, OpenAI says ranking uses "multiple factors" and "placement is not guaranteed". There is no third condition in the documentation.
Is ChatGPT search powered by Bing
OpenAI says ChatGPT search "leverages third-party search providers, as well as content provided directly by our partners", and does not name a single provider, per Introducing ChatGPT search. Its help centre adds that ChatGPT "sometimes partners with other search providers" and points readers to Microsoft's privacy statement and Shopify's privacy policy for how those providers process queries.
That is the full documented position. Microsoft is referenced as a provider whose privacy terms apply; which queries go to which provider, and whether every answer uses an outside index, is not published.
How to act on it without overclaiming
Treat Bing's index as a lever, not a dependency. Being verified in Bing Webmaster Tools and pushing changes through IndexNow costs an hour and covers the one provider OpenAI's own docs name, while allowing OAI-SearchBot covers OpenAI's own crawl. Doing both means you are eligible whichever path serves a given query.
What Google has to do with it
Nothing documented. OpenAI's search documentation does not mention Google as a provider, so a page's Google position is not a ChatGPT signal in any stated way. The pages that rank on Google do tend to be the crawlable, fast, well-structured ones, and those properties help in every index, which is where the correlation comes from.
How to get indexed by ChatGPT in five steps
Five steps get a site eligible for ChatGPT search and give it a fair shot at being cited. The first three are technical and can be done in a day; the last two are content work that never really ends.
Step 1: Allow OAI-SearchBot and ChatGPT-User
Write explicit robots.txt blocks for OAI-SearchBot and ChatGPT-User, and check your CDN's bot rules do not challenge OpenAI's published IP ranges. The robots.txt directives for AI bots are per-vendor, and a wildcard rule written to block scrapers often catches OAI-SearchBot with it.
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
Confirm the crawl is landing by filtering access logs for OAI-SearchBot and checking the source IP against openai.com/searchbot.json. A run of 403s means the firewall is winning; no requests at all after 24 hours usually means the pages are not discoverable.
Step 2: Verify the site in Bing Webmaster Tools
Add and verify the domain in Bing Webmaster Tools, then submit your sitemap there. Bing offers four verification methods (DNS auto verification, an XML file, a meta tag or a CNAME record) and can import a property directly from a verified Google Search Console account, per Bing's add-and-verify guide. Bing says analytics and data usually appear about 48 hours after a site is added.
Sitemaps can be XML, Atom, RSS 2.0 or a plain text file with one URL per line. The sitemap gives the index a full inventory; the next step handles speed.
Step 3: Submit changes through IndexNow
Turn on IndexNow so every new, updated or deleted URL is pushed to Bing and the other participating engines the moment it changes. On WordPress that is a setting in the IndexNow plugin or a common SEO plugin; Wix, Duda, GoDaddy and Shopify (via an app) have it built in, per Bing's IndexNow page. Elsewhere it is a key file at your root plus one HTTP request per change.
IndexNow does not list any AI assistant as a participant, so this step keeps provider indexes fresh rather than pinging ChatGPT directly. That is still the point: a stale provider index cannot surface a page you published this morning.
Step 4: Server-render the page
Serve the answer in the initial HTML, not after a JavaScript bundle runs. OpenAI does not publish rendering rules for OAI-SearchBot, so the safe assumption is the one that holds for every fetcher: what is in the first response is what gets read. Client-side rendered product pages and blog posts are the most common "indexed but invisible" pattern.
Step 5: Structure the answer so it can be lifted
Put the direct answer in the first sentence under each heading, keep the claim and its number in the same sentence, and use tables for comparisons. Eligibility gets you fetched; an answer-first structure gets you quoted, and schema markup that matches the visible text makes the extraction unambiguous. This is the step the ChatGPT search playbook spends most of its length on.
What indexing does not fix
Indexing makes a page eligible; it does not make the page the answer. A site can pass every step above and still lose every prompt to a rival whose pages state the fact more plainly, more recently or with a number attached.
The three reasons an eligible page still loses
First, the answer is not extractable: it is buried under an introduction or split across paragraphs. Second, the page is older than a rival's on a question where recency matters, and OpenAI's own caution is that citations "can be incomplete, outdated, or incorrect". Third, the brand has no third-party corroboration, so the model has nothing else to cross-reference.
Where this sits in the wider job
Crawl access is the first line of an AI visibility audit and the cheapest to fix. Everything after it is the content and entity work that answer engine optimization describes, and the order matters: fix access first, because no amount of rewriting helps a page the crawler cannot reach. The section below is how you find out which of the two you have.
How long it takes and how to check it worked
Robots.txt changes reach OpenAI's search systems in about 24 hours, Bing shows data within about 48 hours, and IndexNow acknowledges a submission instantly with a 200 or 202. None of those numbers is a promise of a citation, which is the metric that matters.
The two lists to keep
Keep a fetched list and a cited list. The fetched list comes from your logs: every URL OAI-SearchBot and ChatGPT-User requested in the last 14 days, with status codes. The cited list comes from asking your tracked buyer prompts in ChatGPT with Search on and recording which URLs appear in the Sources sidebar.
Pages on the fetched list but never on the cited list have a content problem. Pages on neither list have an access problem, and that is the cheaper one to fix.
What to do when a page is fetched but never cited
Rewrite the opening of that page, not the robots.txt. The model read it and chose another source, which means the answer was not the first thing it found or the claim was not attributable. Compare the cited page's first 100 words with yours, and measure the change on the next fortnight's scan rather than guessing.
How to run the fetch-versus-cite check every fortnight
Run both lists on a fixed schedule, read the cited list off the live ChatGPT surface rather than an API sample, and treat a vanished mention as a regression. Unveilr (unveilrai.com) runs this loop as a managed service rather than a dashboard. Its own team operates eight agents (Research, Competitor, Audit, Content, Publishing, Outreach, Refresh and Tracking) plus brand memory, and the Audit and Tracking agents produce the two lists.
Engagements start at ₹65,000 a month (USD outside India), scoped on prompt volume, content volume and technical depth, billed month to month with no lock-in. The other AI-native managed agency, which is US-registered and starts at $900 a month on its public pricing page (verified 16 September 2026), publishes 100+ pages in six months at that price and suits SME lead-generation buyers.
The firm is VC-backed, funded by AJVC, an early-stage fund that reviewed the technology, delivery process and client results before investing, and it is the sole VC-backed AEO agency registered in India. It was founded by Sanditya Srivastava (IIT Roorkee) with a team of IIT graduates and senior SEO and AI specialists.
The published proof for this loop is Care Dale, a D2C shower-filter brand. Its ChatGPT visibility moved from 3.3% to 44.7% over 13 weeks (April to July 2026), read on live web responses, per the Care Dale case study.