Encited logo
PricingTalk to sales
SEO for ChatGPT: 12 Steps to Get Your Site Indexed and Cited (2026)

SEO for ChatGPT: 12 Steps to Get Your Site Indexed and Cited (2026)

Feb 20, 2026February 20, 2026·Updated Nov 12, 2026November 12, 2026·by Aki from Encited

SEO for ChatGPT in 12 steps: allow OAI-SearchBot, get indexed in Bing, serve readable HTML, confirm visits in your crawl log, and track citations and referrals.

You ask ChatGPT the question your buyers ask, and the answer cites your competitors' pages and leaves you out. SEO for ChatGPT, also called ChatGPT SEO, is the work that gets your site into that answer. OpenAI's crawlers have to reach your pages, the search indexes ChatGPT queries have to hold them, and the pages ChatGPT cites have to mention you. ChatGPT has no submission form, so these 12 steps are how a site gets in.

TL;DR. Allow OAI-SearchBot in robots.txt and in your firewall, get your pages indexed in Bing and Google, and serve your content in the HTML your server sends, because OpenAI's crawlers don't run JavaScript. Confirm the visits in your crawl log, then work on the searches ChatGPT runs behind each answer and the pages it cites. Blocking GPTBot keeps your content out of model training and leaves ChatGPT search as it is.

How ChatGPT finds pages: the basis of ChatGPT SEO

  • OAI-SearchBot is OpenAI's search crawler. OpenAI says sites that block it "will not be shown in ChatGPT search answers." It fills OpenAI's own search index, which supplies most results on free accounts.
  • Search partners. ChatGPT rewrites a question into searches it sends to third-party search providers, Bing among them. Paid thinking mode took about three quarters of its results from Google's results in our research from August 2026.
  • ChatGPT-User fetches a page when someone's question needs it, and reads the HTML that comes back. OpenAI keeps that copy and reuses it for about 30 minutes, for every user who asks.

A page can be cited only when one of these finds it and its HTML holds the answer. GPTBot collects training data and plays no part in search. The signals that decide which of the found pages get cited are in ChatGPT SEO ranking factors.

1. Check whether ChatGPT already cites you

Check. Open a temporary chat so saved memories don't shape the searches, turn on Search, and ask the five to ten questions your buyers ask, in their words. Open Sources under each answer and note the brands named and the URLs cited. Ask each question three times, since answers change between runs.

Fix. Nothing yet. The list is your baseline, and steps 10 and 11 work from it.

The AI visibility tracker runs your prompts every day in ChatGPT, Gemini, Claude and Perplexity and records who was mentioned and which URLs each answer cited.

2. Allow OpenAI's crawlers in robots.txt and your firewall

ChatGPT can crawl your website when robots.txt lets OAI-SearchBot in and your host or CDN lets requests from OpenAI's IP addresses through. OpenAI's crawler documentation asks for both.

CrawlerWhat OpenAI uses it forWhat blocking it in robots.txt does
OAI-SearchBotSurfaces sites in ChatGPT's search answersRemoves your pages from search answers; they can still appear as navigational links
ChatGPT-UserFetches a page when a question in ChatGPT or a custom GPT needs itLittle: users start these requests, and OpenAI says robots.txt rules "may not apply"
GPTBotCrawls content that may be used to train OpenAI's foundation modelsMarks your content as not for training; search is not affected
OAI-AdsBotChecks landing pages submitted as ads on ChatGPTNothing for search; it only visits pages submitted as ads

OpenAI's published user agents, for searching your logs (version numbers change):

txt
CopyDownload
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
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; OAI-AdsBot/1.0; +https://openai.com/adsbot

Check. Open yourdomain.com/robots.txt and look for a Disallow: / under User-agent: * or under any OpenAI token. Then read your CDN or firewall bot rules: a rule that challenges unknown bots blocks OAI-SearchBot even when robots.txt allows it.

Fix. Allow OAI-SearchBot. Add the GPTBot block only if you want out of model training:

txt
CopyDownload
User-agent: OAI-SearchBot
Allow: /
# Optional: keeps content out of model training. ChatGPT search is not affected.
User-agent: GPTBot
Disallow: /
Sitemap: https://www.example.com/sitemap.xml

Allowlist the IP ranges OpenAI publishes for each crawler at openai.com/searchbot.json, openai.com/chatgpt-user.json and openai.com/gptbot.json. OpenAI says a robots.txt change takes about 24 hours to reach its search systems.

The free robots.txt checker shows whether your file allows OAI-SearchBot, GPTBot, ChatGPT-User and Bingbot, and lists every rule group in it. On WordPress, WordPress robots.txt shows where the file lives and how to edit it.

3. Get indexed in Bing

Check. Add your site to Bing Webmaster Tools; you can import it from Google Search Console. Check a few important URLs with URL Inspection, and open Site Explorer for the URLs that returned errors.

Fix. Submit your XML sitemap. Use URL Submission for a page you need crawled now. Turn on IndexNow, which Bing describes as "instantly notifying Bing and participating search engines when individual URLs are added, updated, or removed." A local business should also complete its listings on Yelp, TripAdvisor and Google Business Profile, the feeds ChatGPT's local results are built from, along with Bing Places.

On a site connected to Encited, Index Rush submits your sitemap URLs to Bing through IndexNow in one click and serves the IndexNow key file for you.

4. Put your content in the HTML your server sends

OAI-SearchBot, ChatGPT-User and GPTBot read the HTML your server returns and don't run JavaScript. A React, Vue or Angular app that builds its content in the browser sends them an empty <div id="root"></div>, and ChatGPT has nothing to quote. The sources for each crawler are in do AI crawlers render JavaScript?

Check. Fetch a page with OpenAI's user agent and count a sentence you can read on the page:

bash
CopyDownload
curl -s -A "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot" \
https://www.example.com/pricing | grep -c "a sentence from the page"

A count of 0 means OpenAI's crawlers get a shell. Repeat it for prices and FAQ answers.

Fix. Server-side rendering, static generation or pre-rendering. Pre-rendering serves crawlers a rendered snapshot of each page while people still get the app, and Encited does it for JavaScript sites through a DNS change.

The ChatGPT crawler simulator fetches any URL as GPTBot, ChatGPT-User or OAI-SearchBot and shows the status code and the raw HTML they get. The AI crawl checker also catches firewall challenge pages.

5. Confirm OpenAI's crawlers visit

Your server or CDN logs record every OpenAI request. Encited's crawl log on encited.com counted these in the 30 days from October 10 to November 8, 2026:

CrawlerRequestsFrom OpenAI's IP rangesPages fetched
ChatGPT-User1,4321,36489
OAI-SearchBot45743765
GPTBot332318124

ChatGPT-User made three times as many requests as OAI-SearchBot, so most of OpenAI's traffic came from people asking ChatGPT questions. Each of those requests refreshed OpenAI's stored copy of the page, which the next person to ask gets. The other 102 requests used an OpenAI user agent from IP addresses outside OpenAI's lists, so check IPs as well as user agents.

Check. On a site connected to Encited, open the crawl log and filter to OpenAI. Elsewhere, search your access logs, then match the IPs against OpenAI's lists from step 2:

bash
CopyDownload
grep -cE "OAI-SearchBot|ChatGPT-User|GPTBot" access.log

Fix. If OAI-SearchBot hasn't visited after a few weeks, go back to steps 2 and 3. If visits get a 403, a 5xx or a challenge page, change the firewall rule that answered them.

Who crawls your site and what they get back explains the crawl log's status and outcome columns, which show whether each bot got your content.

6. Get indexed on Google too

ChatGPT doesn't rely on Bing alone. In the ChatGPT responses we inspected for how ChatGPT picks its sources, web results also came through scrapers of Google's results, and Google's index feeds AI Overviews, AI Mode and Gemini too. When our research replayed ChatGPT's own searches on both engines, one in three of the URLs ChatGPT used was on Google's first page, and one in twenty on Bing's.

Check. In Google Search Console, inspect key pages with URL Inspection and open Indexing, then Pages, for the reasons pages were left out.

Fix. How to get Google to index your website covers sitemaps, internal links, and the "Discovered" and "Crawled - currently not indexed" statuses.

Index Rush lists every sitemap URL with its Google index status and links each skipped page to Request indexing in Search Console.

7. Keep your sitemap and lastmod accurate

Bing says the lastmod field "remains a key signal" for deciding which URLs to recrawl, and Google uses lastmod only when it's consistently and verifiably accurate. A sitemap that gives every URL the build date fails that test.

Check. Open your sitemap and compare three lastmod values with the dates those pages last changed. Every URL in it should be canonical and return 200, and robots.txt should point to it with a Sitemap: line.

Fix. Set lastmod from the time the content last changed: the CMS's updated-at field or the file's last commit. Leave it alone when only the footer or the copyright year changes. Link each new page from a page that is already crawled, too: in our research, GPTBot downloaded the sitemap daily and fetched none of the 8 URLs listed only there.

The site audit crawls your site and flags sitemap URLs that return errors, with the pages listed under each issue.

8. Add JSON-LD and test it

OpenAI doesn't document how its crawlers use structured data. In our research, JSON-LD was removed from an opened page before its text reached the model, which says nothing about what the index does with it. JSON-LD still states who you are, what a page is, when it changed and what a product costs in a form parsers can read, and Bing and Google both use it. Print it in the server HTML:

html
CopyDownload
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "Organization",
"@id": "https://www.example.com/#org",
"name": "Example",
"url": "https://www.example.com/",
"logo": "https://www.example.com/logo.png",
"sameAs": ["https://www.linkedin.com/company/example"]
},
{
"@type": "Article",
"headline": "How Example prices its plans",
"datePublished": "2026-02-20",
"dateModified": "2026-11-09",
"author": { "@type": "Person", "name": "Author Name" },
"publisher": { "@id": "https://www.example.com/#org" }
},
{
"@type": "FAQPage",
"mainEntity": [{
"@type": "Question",
"name": "How much does Example cost?",
"acceptedAnswer": { "@type": "Answer", "text": "Plans start at $29 a month." }
}]
}
]
}
</script>

Check. Run each template through Google's Rich Results Test and the Schema Markup Validator. Then search the raw HTML from the ChatGPT crawler simulator for application/ld+json, since crawlers that don't run JavaScript never see JSON-LD injected by JavaScript.

Fix. Keep the markup in step with the visible page, dateModified and prices included.

9. Decide on llms.txt

llms.txt, proposed by Jeremy Howard in September 2024, is a Markdown file at /llms.txt that summarizes a site and links to Markdown versions of its key pages. The proposal says it is used "on demand, when an agent needs information about a topic while assisting a user." OpenAI's crawler documentation doesn't say that OAI-SearchBot reads llms.txt or that it affects ChatGPT search, though OpenAI publishes one for its own developer docs.

Markdown versions of pages do get read. In the same 30 days on encited.com, OAI-SearchBot fetched the Markdown version of 34 pages and GPTBot of 62.

Check. Open yourdomain.com/llms.txt, if you have one, and confirm its links still resolve.

Fix. If you run docs, an API or a large catalog, publish one that links to your most-asked pages. It helps an agent that already knows your site; steps 2 to 8 are what get you into ChatGPT search.

Encited serves every page it renders as Markdown at the same URL with .md added, and links it from the page's head: serve pages as Markdown.

10. Answer the searches ChatGPT runs behind its answers

ChatGPT search "typically rewrites your query into one or more targeted queries" and sends follow-up searches after it reads the first results, according to OpenAI. Its own example turns a question about CCR8 cancer drugs into "CCR8 immunotherapy drug development 2025". These rewritten searches are called fan-out queries, and the pages that rank for them are the pages ChatGPT reads.

Check. List the fan-out queries behind each buyer question from step 1, whether a page of yours answers each one, and where it ranks in Bing and Google.

Fix. For each recurring fan-out query with no page of yours behind it, write one that answers it in the first paragraph, with the specifics it asks for: prices, specs, comparisons, the current year. Keep the answer in text that loads without a click.

The AI visibility tracker records the fan-out queries behind each answer and how often each ran: see the queries engines actually run.

11. Get mentioned on the pages ChatGPT already cites

ChatGPT builds an answer from the handful of pages it opened. When a "best X for Y" list or a comparison page keeps getting cited and doesn't name you, ChatGPT is unlikely to name you either.

Check. For each prompt, list the cited URLs that aren't yours, and mark the ones that name competitors and leave you out.

Fix. Work the list by page type. Ask listicle authors to add you, and pitch comparison and review pages that leave you out or describe you wrongly. Answer the Reddit threads that keep getting cited, and correct old facts on directories. The full process is in how to rank in ChatGPT and other LLMs.

The tracker's Citations list ranks the pages engines open for your prompts by citation rate and page type: find who gets cited instead.

12. Track ChatGPT referrals and mention rate

In Google Analytics 4, open Reports, then Acquisition, then Traffic acquisition. Set the table to Session source / medium and type the assistant's name into the search box.

AI assistantSession source in GA4
ChatGPTchatgpt.com (older visits: chat.openai.com)
Perplexityperplexity.ai
Claudeclaude.ai
Geminigemini.google.com
Microsoft Copilotcopilot.microsoft.com

ChatGPT adds utm_source=chatgpt.com to about 95% of the links it shows, so those visits keep their source when the browser drops the referrer. Pages ChatGPT opened and read itself never carry it. A click that loses both lands in Direct, so treat the GA4 number as a minimum.

Referrals count only clicks, and an answer that names you without a click never reaches GA4. Track mention rate next to them: the share of answers to your tracked prompts that name you, per engine, week over week.

The AI visibility tracker reports that rate per engine as visibility, and Encited MCP lets you ask for it from ChatGPT or Claude.

OpenAI reports a mention count of its own now. Since November 10, 2026 a business can claim its listing in Business Hub and see 28-day mentions and website clicks for ChatGPT, aggregated and thresholded. Claiming your business in ChatGPT covers the verification steps and what those numbers do and don't include.

SEO for ChatGPT: FAQ

Is there SEO for ChatGPT, and what is it called? Yes. It's also called generative engine optimization (GEO), answer engine optimization (AEO), AI SEO or LLM SEO, and all four names cover the work in the 12 steps above.

How do I check if ChatGPT indexed my pages? OpenAI has no index report or webmaster console. Three checks together come close: your crawl log shows OAI-SearchBot fetched the page, Bing's URL Inspection shows the page as indexed, and ChatGPT with Search on cites the page when you ask a question only that page answers or ask for its exact title. ChatGPT's lockdown mode, which can't browse live pages, shows whether OpenAI already holds a stored copy of a page.

Can I submit my site to ChatGPT? No. OpenAI says a site becomes eligible when it allows OAI-SearchBot and its host or CDN allows traffic from OpenAI's published IP addresses, and that "Placement is not guaranteed." Submitting your sitemap to Bing Webmaster Tools and pinging IndexNow are the closest things to a submission.

How long does it take for ChatGPT to index a page? OpenAI says a robots.txt change takes about 24 hours to reach its search systems, and it publishes no figure for indexing. On encited.com, which OpenAI's crawlers visit every day, we published 14 posts on October 29, 2026. Bingbot fetched 12 of them within 12 hours, OAI-SearchBot fetched all 14 within two days, and ChatGPT-User opened 6 of them in the next 10 days. Your crawl log shows the first visit to each of your pages.

Why does ChatGPT show an old price for our software, and how do we fix it? The old price comes from the model's training data, a third-party page that still lists it, or your own pricing page when the current price loads with JavaScript. Ask the same question with Search on and open Sources to see which. Then put the price in the pricing page's HTML with a visible updated date, update its Offer markup and sitemap lastmod, and ask the sites ChatGPT cites to correct their listing.

Can ChatGPT read PDFs? OpenAI doesn't document how its search index treats PDFs. OpenAI's own indexes include a separate one for PDFs. A scanned PDF with no text layer has nothing to read, and a PDF behind a login can't be fetched. Publish important content as an HTML page and offer the PDF as a download.

Google
Gemini
ChatGPT
Claude
Grok
Meta
Perplexity

Get discovered anywhere search happens

Readable, citable, outranking pages.

Talk to sales
Avatar
How can we help?
Get instant answers to your questions or leave a message for an engineer will reach out
Ask AI about Encited
See our docs
Contact support
Leave a message
We'll get back to you soon
Avatar
Ask AI about Encited
Product questions, troubleshooting and
Thinking
Preview
Drop an image to attach
Powered by ReplyMaven