On this page · 8 sections
Summary. To rank your website on ChatGPT, let OpenAI's OAI-SearchBot crawler in through robots.txt and your CDN, get your pages indexed by the search providers ChatGPT draws on (OpenAI points to Microsoft, which runs Bing), and publish pages that answer specific buyer questions with facts other sites lack. OpenAI says placement is not guaranteed and ads do not influence ChatGPT's answers, so there is no paid route into organic results. Guides often get the crawler wrong: of OpenAI's four agents, only OAI-SearchBot decides whether you appear in ChatGPT search, GPTBot collects training data, and robots.txt changes take about 24 hours to reach OpenAI's search systems. On 10 November 2026 OpenAI opened Business Hub in beta, where a verified business can update its details, add product feeds and see aggregated mentions and website clicks. ChatGPT shopping is still live only in the US. Beyond ChatGPT, Bing has reported AI citations in Webmaster Tools since 10 February 2026, and Google says its AI features need no special files or schema. This guide gives 10 steps, a measurement plan and the tactics that do not work.
About this article. For marketing leads, founders and IT teams who want their site cited in ChatGPT and other AI assistants. We checked crawler roles, user-agent versions (OAI-SearchBot/1.4, GPTBot/1.4, ChatGPT-User/1.0, OAI-AdsBot/1.0), feed fields, settings and dates against OpenAI's documentation and announcements, Microsoft Bing's blogs, IndexNow.org, Cloudflare's documentation and Google Search Central on 11 November 2026, reading OpenAI pages that refuse automated access through Internet Archive copies. OpenAI does not publish ChatGPT's ranking factors, so we say where advice rests on documentation and where on a study or inference. General information, not legal advice; no vendor paid for inclusion, and nothing here guarantees a citation. Researched with AI assistance and reviewed by our editorial team.
The four OpenAI agents, and the one that decides ChatGPT search
Some AEO checklists treat GPTBot as the crawler that gets you into ChatGPT. OpenAI's crawler page treats each robots.txt setting as independent, and its own example is the setup most businesses want: allow OAI-SearchBot to appear in search, disallow GPTBot to stay out of model training.
| User agent | What OpenAI says it does | If you block it in robots.txt | Published IP list | What to do for visibility |
|---|---|---|---|---|
| OAI-SearchBot/1.4 | Surfaces websites in ChatGPT's search features | Not shown in ChatGPT search answers, though the site can still appear as a navigational link | openai.com/searchbot.json | Allow |
| GPTBot/1.4 | Crawls content that may be used to train OpenAI's foundation models | Signals your content should not be used for training; no stated effect on search | openai.com/gptbot.json | Your licensing decision |
| ChatGPT-User/1.0 | Visits a page when a person's request in ChatGPT or a custom GPT needs it | OpenAI says robots.txt rules may not apply; not used to decide search inclusion | openai.com/chatgpt-user.json | Do not block at the firewall unless you mean to refuse those requests |
| OAI-AdsBot/1.0 | Checks the safety of pages submitted as ChatGPT ads | Visits only ad landing pages; data not used for training | openai.com/adsbot.json | Allow if you advertise |
OpenAI gives those version numbers as examples that may change, so match on the bot name. A robots.txt that keeps you in ChatGPT search while opting out of training looks like this:
User-agent: OAI-SearchBot
Allow: /
User-agent: GPTBot
Disallow: /
The same page adds three practical points. If you allow both bots, OpenAI may reuse one crawl for both purposes. Its search systems take about 24 hours to pick up a robots.txt change. And ChatGPT-User plays no part in search inclusion, so managing it in robots.txt does not change your search visibility.
To keep a page out of ChatGPT entirely, blocking OAI-SearchBot is not enough, because an opted-out site can still appear as a navigational link. OpenAI's publisher FAQ advised a noindex meta tag on a page the crawler can still fetch. That FAQ sat under the ChatGPT Atlas browser, and its address now leads to an Atlas retirement notice. No current OpenAI page repeats the advice, so treat noindex as the standard web control rather than an OpenAI guarantee, and put anything private behind a login.
This guide focuses on ChatGPT; for Google's AI features, see our guide to ranking in Google AI Overviews.
How ChatGPT search finds and cites pages
OpenAI publishes the inputs, not the ranking formula. It says ChatGPT ranks results on several factors meant to surface relevant, reliable information, and that placement is not guaranteed.
- Search providers. ChatGPT search sometimes rewrites a prompt into targeted queries and sends them to third-party search providers. OpenAI's help article points readers to Microsoft's and Shopify's privacy statements for how those providers handle queries, but does not say how much weight their results carry.
- Its own crawl. OAI-SearchBot fetches pages so they can be surfaced and linked. OpenAI's eligibility advice is to allow OAI-SearchBot and make sure your host or CDN accepts its published IP addresses.
- Citations. Answers that use web search may include citations, and a Sources view lists cited sources plus other relevant links.
- Live fetches. When a person's request needs a specific page, ChatGPT-User may visit it during the conversation.
- Product and business data. Shopping results draw on merchant feeds as well as crawled pages, and OpenAI says it matches products to people using their preferences and past interactions, so two users can see different results. Since 10 November 2026, verified businesses can also supply details and product feeds through Business Hub.
What OpenAI does not publish: ranking factors, whether OAI-SearchBot renders JavaScript, and how it weighs mentions of your brand elsewhere. Independent research fills some of the gap. Suganthan Mohanadasan captured the data behind conversations on one ChatGPT Plus account in July 2026: across 57 conversations, only 110 of 3,554 retrieved pages (3.1%) earned a citation, and brands that ChatGPT wrote into its own search queries reached the answer 68.9% of the time, against 2.1% for brands whose pages were only fetched. He calls every percentage directional, because it all comes from a single account. The direction is still useful: being retrieved is not the same as being cited, and being known as an answer to the question helps.
Step by step: how to get your website cited by ChatGPT
These 10 steps run from access to content to measurement. The first three are technical and quick; the rest are ongoing work.
1. Open the door to OAI-SearchBot at every layer
Check robots.txt for any rule that blocks OAI-SearchBot, including a blanket disallow under User-agent: *. Then check your CDN and firewall, because an edge block stops OpenAI reading the page whatever robots.txt says. Cloudflare's AI Crawler and Agent access policies, for example, allow or block Search, Agent and Training traffic separately; new domains start with all three allowed, but an older zone or custom rule may differ. Allowlist by OpenAI's published IP ranges, as OpenAI asks, rather than by user-agent string, which anyone can fake. Allow about 24 hours after a change, then look for OAI-SearchBot requests returning 200 in your logs.
2. Get indexed by Bing as well as Google
OpenAI's help article points to Microsoft as one of ChatGPT's search providers, so a page Bing has not indexed starts at a disadvantage. Verify your site in Bing Webmaster Tools and submit your XML sitemap there, or list it in robots.txt; Bing says it fetches a submitted sitemap straight away, then typically rechecks it at least daily. Our guide to ranking on Bing covers the set-up. Then add IndexNow, which Microsoft recommends for keeping AI answers current. You host a key of 8 to 128 characters at your site root and can submit up to 10,000 URLs in one POST request; a 200 response means the URLs were received, not indexed. IndexNow recommends automating submissions whenever pages are added, updated or deleted.
3. Put the answer in the HTML
OpenAI does not say whether OAI-SearchBot executes JavaScript, so do not rely on it. Server-render or pre-render the main content, prices, specifications and FAQs so they appear in the HTML response. Google's guidance makes a related point: make sure important content is available as text. Test by fetching a page with JavaScript switched off; whatever is missing may be invisible to AI crawlers.
4. Answer the questions buyers actually ask
People ask ChatGPT full questions, such as which payroll software suits a 50-person company in Pune. Build pages that answer those questions directly: use the question as a heading, answer it in the first two sentences, then give the detail. Microsoft's advice in its AI Performance announcement is unusually specific: deepen expertise, use clear headings, tables and FAQ sections, back claims with examples, data and cited sources, keep content fresh, and make text, images and video describe the same products and entities.
5. Publish facts other sites do not have
An answer engine has little reason to cite a page that repeats what ten others say. Google's AI search guide makes the same case: unique, first-hand content is likely to matter more than any other tactic, and mass-producing pages for every query variation can breach its scaled content abuse policy. Prices, specifications, limits, delivery areas, compatibility tables, original survey results and dated changelogs give an assistant something to quote. Name the author or team, date the page and cite your sources. Never invent statistics or customer stories to look citable.
6. Be described consistently across the web
Assistants learn who you are from more than your own site. Use the same company name, one-line description, address and product names on your website, Google Business Profile, Bing Places, review sites, marketplace listings and industry directories, and earn mentions in independent sources your buyers read. Do not manufacture them: Google says chasing inauthentic mentions is less helpful than it looks, and its spam systems apply to AI features too. OpenAI does not say how it weighs these signals, so treat this as good practice supported by research such as the study above, not as a documented ranking factor.
7. Claim your business and share a product feed
Business Hub, opened in beta on 10 November 2026, lets a business claim its listing in ChatGPT and, once verified, update key details and add product feeds with pricing and availability. OpenAI says any eligible business located where ChatGPT is available can claim, and Shopify merchants can connect their store to speed up verification. Search Engine Roundtable reported verification by phone call, text message or DNS record. OpenAI's merchants page calls a feed optional if ChatGPT already crawls your site, but says it gives more control over how products appear; Shopify and Etsy catalogues are already integrated. Its feed reference lists 9 required fields per item (item_id, title, description, url, brand, seller_name, image_url, availability and price), suggests titles of at most 150 characters and turns search eligibility on by default. Shopping is live only for US users and the standard feed upload targets the US, so Indian and other non-US merchants should claim their business and prepare clean data now, but expect product results to reach their buyers later.
8. Keep local and service details current
For location questions, assistants need accurate addresses, hours and contact details. Microsoft recommends registering with Bing Places for Business so those details stay current and eligible for AI answers. Keep Google Business Profile and your ChatGPT business profile in step with it; Google says Business Profile and Merchant Center help products and services show up in its AI responses. Service businesses should publish a clear page per service with scope, process, typical timelines and price ranges where they can share them.
9. Use structured data that matches the page
Organization, Product, Article, FAQPage and LocalBusiness markup help machines read your pages, provided the markup matches the visible text. Google says structured data is not required for its generative AI features and there is no special schema.org markup to add, though it still helps with rich results. OpenAI publishes no structured-data guidance for ChatGPT search. Use schema for accuracy, not as a ranking trick.
10. Refresh what gets cited
Find the pages AI systems already use and improve them first. Bing's AI Performance report shows page-level citation counts and a sample of grounding queries for Microsoft's AI surfaces, and since June 2026 it also groups queries into intents and topics and shows citation share. Business Hub shows discovery topics, the themes of conversations where your website is cited. Your logs show which pages ChatGPT-User opens. Update those pages when prices, specifications or facts change, change the date only when the content changes, and resubmit them through IndexNow.
How to measure your ChatGPT and AI search visibility
OpenAI's Business Hub beta, opened on 10 November 2026, shows a verified business aggregated counts of mentions and website clicks, plus discovery topics. OpenAI does not share users' conversations, and insights appear only once activity passes its thresholds, so small sites may see little at first. Search Engine Roundtable reported that the dashboard's help text says a recorded click does not confirm a visit or conversion. Our Business Hub explainer covers the background. OpenAI's retired publisher FAQ said ChatGPT adds utm_source=chatgpt.com to referral links, so confirm the tag in your own analytics. Combine these signals.
| Signal | Where to see it | What it tells you | Limit |
|---|---|---|---|
| OAI-SearchBot requests | Server or CDN logs, checked against searchbot.json | Pages are being fetched for ChatGPT search | A crawl is not a citation |
| ChatGPT-User requests | Server or CDN logs | A conversation caused ChatGPT to open your page | Not every fetch ends up cited |
| utm_source=chatgpt.com sessions | GA4, filtered by session source | Clicks from ChatGPT answers to your site | Undercounts; clicks without the tag land in direct |
| Mentions, clicks and discovery topics | ChatGPT Business Hub (beta), for verified businesses | How often ChatGPT names you and records clicks to your site | Aggregated, shown only above thresholds, no conversations |
| Citations and grounding queries | Bing Webmaster Tools, AI Performance report | Citations in Copilot, Bing AI summaries and select partners | Microsoft does not say ChatGPT is a partner; queries are sampled |
| Monthly prompt panel | Spreadsheet, or a paid tracker | Whether you are named or linked for target questions | A sample of prompts, not every user |
Start with a baseline. Record 90 days of chatgpt.com referrals and OpenAI bot counts, and run 20 to 30 buyer-intent prompts through ChatGPT once a month, noting whether you are named, whether you are linked and who else appears. Re-run on the same day each month so the numbers compare. Our GEO measurement guide covers the GA4 channel set-up and compares paid trackers. Judge progress by month-on-month change on your own prompts; there is no universal benchmark for a good citation rate.
What does not work
These are the common mistakes.
- Allowing GPTBot and assuming that covers ChatGPT search. GPTBot is for training. OAI-SearchBot is the search crawler.
- Blocking OAI-SearchBot and assuming the page disappears. ChatGPT can still show the site as a navigational link. Use noindex on a crawlable page, and a login for anything private.
- Using robots.txt to control ChatGPT-User. OpenAI says robots.txt rules may not apply to fetches a person starts. robots.txt is not access control.
- Buying ads to lift organic answers. OpenAI's advertising principles say ads do not influence ChatGPT's answers and are always labelled and kept separate. Sponsored Business Agent placements are labelled and separate too. Ads may be worth testing on their own merits, but they will not change organic citations.
- Treating llms.txt as a ranking lever. OpenAI's crawler documentation does not mention llms.txt, and Google says Search ignores such files. Our llms.txt explainer covers where the file can still help, such as developer documentation.
- Paying for a submit-your-site-to-ChatGPT service. OpenAI documents no URL submission tool for organic ChatGPT search. The routes it documents are crawling, Business Hub, product feeds and paid ads.
- Rewriting everything into chunks or adding AI schema. Google says neither is needed, and OpenAI asks for neither. Clear, complete pages for people are the target.
- Chasing a fixed rank. ChatGPT's answers vary by conversation, and OpenAI says product results are personalised. Track whether you appear across a panel of questions, not a position for one.
- Astroturfing forums and review sites. Fake posts and reviews break those platforms' rules and put your accounts and reputation at risk.
For IT teams and businesses
Getting cited by ChatGPT is partly an infrastructure job, and that is where accidental blocks often happen.
Bot management. Write an explicit policy for each OpenAI agent and apply it at the CDN or WAF as well as in robots.txt. Allowlist OAI-SearchBot and ChatGPT-User by their published IP lists, rate-limit rather than block if load is a concern, and treat GPTBot as a licensing decision owned by whoever owns the content. Our crawl-to-refer analysis sets out how to split training crawls from retrieval and agent fetches.
Change control. Version robots.txt and bot rules, review changes, and alert on unexpected drops in OAI-SearchBot or Bingbot traffic. Make sure staging is blocked and production is not, a mix-up that is easy to ship during a migration.
Logs. Keep raw access logs, or at least daily counts by agent, for 13 months or more so you can compare year on year.
Business Hub access. Treat the claimed ChatGPT profile like any brand account: verify it through a company-controlled number or DNS record, and record who holds access.
Data protection. Anything a crawler can reach is public. Under India's DPDP Act 2023 and the GDPR, pages that expose personal data are a problem whoever's bot reads them, so review what sits on crawlable URLs, and keep customer data out of prompts sent to AI visibility trackers.
India. Indian businesses can be cited in organic ChatGPT answers today, and OpenAI says any eligible business located where ChatGPT is available can claim a Business Hub profile. Shopping, Business Agents and the standard feed upload are US-first, so prioritise steps 1 to 6 and 10, which work in every market.
How eCorpIT can help
eCorpIT is a Gurugram-based technology organisation, founded in 2021, assessed at CMMI Level 5 and MSME certified, with senior-led engineering teams working across AWS, Microsoft and Google platforms. Our SEO, AEO and GEO audit checks each layer that decides whether ChatGPT can cite you, from robots.txt and CDN bot rules to Bing indexing, rendering, structured data and the measurement plan above. Talk to us at /contact-us/.
Last updated: 11 November 2026.
