“ChatGPT SEO” means two different things, and the search results mix them. Most of the pages that rank, from Ahrefs, NoGood and SimpleTiger among them, are about using ChatGPT to do SEO work: prompts for keywords, titles, schema and audits. A smaller group, Semrush’s guide for one, is about getting your pages cited when ChatGPT searches the web. This guide covers both, in that order.
When we read the top results for “chatgpt seo” on 11 November 2026, the prompt guides were up to three years old and none cited OpenAI’s own documentation. The citation guides gave advice without saying what OpenAI states and what is a guess. Below, every claim about how ChatGPT search works links to OpenAI’s own pages, read on 11 November 2026, and the prompts are ones you can paste today.
How do you use ChatGPT for SEO work?
What is ChatGPT good at in SEO, and what is it bad at?
ChatGPT is a language model. It is good at turning one text into another: a keyword list into clusters, a page into a meta description, a spec into JSON-LD. It is bad at anything that needs data it cannot see.
| Task | Use ChatGPT? | Why |
|---|---|---|
| Title tags, meta descriptions, headline variants | Yes | Short text with clear rules (length, keyword, promise) |
| Schema markup (JSON-LD), hreflang tags, redirect rules, robots.txt drafts | Yes, then validate | Pattern work; run the output through a validator before you ship it |
| Regex for Search Console or GA4 filters | Yes, then test | Fast to write, easy to check on a sample |
| Grouping a keyword list you paste in | Yes | Sorting text by meaning is a language task |
| Outlines from a SERP you summarize | Yes | Turns your notes on the top 5 into a structure |
| Keyword research from scratch | No | It has no search volume or difficulty data; use a keyword tool and paste the results in |
| Facts, statistics and prices | No | It can state wrong numbers with confidence; source every number yourself |
| Full articles published as written | No | Output tends to be generic, and publishing many unedited pages risks Google’s scaled content policy |
| Technical audits of a live site | Only with data | It cannot crawl your site in the default chat; paste crawl exports or page source |
Ahrefs’ own test of ChatGPT for SEO reached a similar split: it rated regex, code snippets, titles and outlines as good uses, and full content creation, keyword research and search intent classification as weak ones (Ahrefs).
10 ChatGPT prompts for SEO tasks
Each prompt asks for one output, gives the rules, and asks ChatGPT to flag what it is unsure of. Replace the brackets.
1. Group a keyword list into page clusters
Here are 120 keywords with monthly US search volume, exported from my keyword tool: [paste]. Group them into clusters where one page could rank for every keyword in the cluster. For each cluster give a page title, the primary keyword (highest volume), and the search intent: informational, commercial, transactional or navigational. Do not add keywords that are not in my list. Mark any keyword whose intent you are unsure of.
2. Write title tags and meta descriptions
Page: [URL and one-paragraph summary]. Primary keyword: [keyword]. Write 5 title tags of 60 characters or fewer that start with the keyword or put it in the first half, and 5 meta descriptions of 155 characters or fewer that state what the reader gets. Count the characters and show the count for each. No clickbait, no year unless the page is dated.
3. Outline a page from the top 5 results
I read the top 5 Google results for [keyword]. Here is what each covers and what each misses: [your notes]. Write an H2/H3 outline for a page that covers everything the top 5 cover, fills the gaps I listed, and answers these People Also Ask questions: [paste]. Phrase each H2 as the question a reader would ask.
4. Generate schema markup
Write JSON-LD for an FAQPage with these questions and answers: [paste]. Use only properties defined on schema.org for FAQPage, Question and Answer. Output the script tag only.
5. Write a regex filter for Search Console
Write a regex for Google Search Console’s query filter that matches queries containing any of these words: pricing, price, cost, costs, how much. Search Console uses RE2 syntax. Then list 5 queries it should match and 3 it should not, so I can test it.
6. Draft hreflang tags
My site has English pages at /en/, German at /de/ and French at /fr/, all on example.com. Write hreflang link tags for the page [path], including x-default pointing to /en/. Explain in one line where they go.
7. Draft redirect rules
I am moving these URLs: [old URL, new URL pairs]. Write 301 redirect rules for [Apache .htaccess / Nginx / Vercel vercel.json]. Flag any pair that would create a redirect chain or loop.
8. Suggest internal links
Here is a list of my published pages with their target keywords: [paste]. Here is a new page: [title and summary]. Suggest 5 existing pages that should link to the new page and 3 the new page should link to, with the anchor text for each. Only use pages from my list.
9. Refresh a page from Search Console data
Here are the queries this page got impressions for in the last 3 months, with clicks, impressions and average position: [paste]. Here is the page: [paste]. List the queries with high impressions and low clicks that the page does not answer directly, and suggest a section or sentence that would answer each.
10. Rewrite for clarity
Rewrite this paragraph so each sentence states one fact, the answer comes in the first sentence, and nothing is lost: [paste]. Keep every number exactly as written.
Prompts 4 to 7 produce code. Test each one before it reaches your live site: Google’s Rich Results Test for schema, a sample of real queries for regex, and a staging deploy for redirects.
What does Google say about content written with ChatGPT?
Google’s position is that it judges the content, not the tool. Its February 2023 guidance says its systems reward “original, high-quality content” and that “using AI doesn’t give content any special gains. It’s just content” (Google Search Central). The same guidance says that using automation “with the primary purpose of manipulating ranking in search results is a violation of our spam policies.”
The line is drawn in Google’s spam policies under scaled content abuse, which includes “using generative AI tools or other similar tools to generate many pages without adding value for users” (Google spam policies). In practice: use ChatGPT for structure, drafts and code, then add what it cannot produce, such as your own data, customer examples, screenshots and checked facts.
How do you get your pages cited in ChatGPT search?
How does ChatGPT search choose its sources?
OpenAI’s help center describes the process in a few steps (OpenAI, searching the web with ChatGPT):
- ChatGPT decides whether to search. It “may search the web automatically when your question would benefit from current information,” or the user turns search on.
- It rewrites the question into search queries. ChatGPT search “sometimes partners with other search providers” and “typically rewrites your query into one or more targeted queries that it sends those providers.” After reading the first results, it may send more specific follow-up queries.
- It ranks and cites. “ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed.” Answers may include citations and a Sources panel.
OpenAI does not name its search providers on that page or list its ranking factors. Its 2024 launch post said search “leverages third-party search providers, as well as content provided directly by our partners” (OpenAI). Anyone who tells you the exact ranking formula is guessing.
What this means for your pages: the follow-up queries are specific and long, so a page that answers one narrow question directly has a better chance of matching one of them than a general page does.
Which OpenAI crawlers should you allow?
OpenAI runs separate user agents, and each one is controlled on its own in robots.txt (OpenAI crawler documentation):
| User agent | What it does | What blocking it does |
|---|---|---|
| OAI-SearchBot | Crawls sites “to surface websites in search results in ChatGPT’s search features” | Your site “will not be shown in ChatGPT search answers,” though it can still appear as a navigational link |
| GPTBot | Crawls content “that may be used in training our generative AI foundation models” | Tells OpenAI not to use your content for training; does not affect search |
| ChatGPT-User | Visits a page when a user’s request in ChatGPT or a custom GPT needs it | OpenAI says robots.txt rules “may not apply” because a user started the visit, and this agent is not used to decide what appears in search |
| OAI-AdsBot | Checks landing pages submitted as ChatGPT ads | Only visits pages submitted as ads |
OpenAI says it can take about 24 hours for its search systems to adjust after you change robots.txt. A robots.txt that keeps a site eligible for ChatGPT search while opting out of training looks like this:
User-agent: OAI-SearchBot
Allow: /
User-agent: GPTBot
Disallow: /
If you want to be found and do not mind training use, allow both. Then check one more layer: OpenAI’s help center says the website host or content delivery network must allow traffic from OpenAI’s published search bot IP addresses. Bot protection on Cloudflare, Vercel or a WAF can block OAI-SearchBot even when robots.txt allows it. Ask whoever runs your hosting to check the firewall logs for OAI-SearchBot requests.
What makes a page more likely to be cited?
OpenAI does not publish a checklist beyond crawl access, so the steps below are practice, not OpenAI rules. They follow from how the process works: specific queries sent to search providers, then a ranked set of sources.
- Be crawlable and indexed. Allow OAI-SearchBot, allow it at the CDN, and make sure your key pages are indexed in the major search engines, since ChatGPT draws on third-party search providers.
- Answer the question in the first sentence under each heading. Phrase headings as the questions buyers ask, and put the answer, with its number and source, right under them. Our answer engine optimization entry covers this format in detail.
- Publish the facts only you have. Pricing, integration lists, implementation time, customer counts you can source. A page that states a fact clearly and first is easier to cite than one that repeats what ten others say.
- Keep your facts the same everywhere. If your pricing page, G2 profile and LinkedIn page give different numbers, an answer may pick the wrong one.
- Earn mentions on the pages that answers already cite. Run your buyer questions, open the Sources panel, and note which review sites, lists and articles appear. Getting listed or quoted there is often faster than ranking a new page.
- Consider an llms.txt file, with low expectations. llms.txt is a proposed standard for pointing AI tools to your key pages. OpenAI’s crawler documentation does not mention it. It is cheap to add; our free llms.txt generator builds one in a minute.
How this differs from ranking on Google, and what stays the same, is laid out in our GEO vs SEO comparison.
How do you measure whether ChatGPT cites you?
Use two measures, because each misses part of the picture.
- Referral traffic. In GA4 or your analytics tool, look for sessions with chatgpt.com as the source. People who read an answer and later type your URL show up as direct traffic, so treat the count as a floor.
- Answer checks. Keep a fixed list of 20 to 50 questions your buyers ask. Run them in ChatGPT with search on once a month, and log whether each answer names you, which competitors it names and which URLs it cites. Answers can change from one run to the next, so run each question at least twice. Tools can automate this; our guide to AI visibility tools compares 10 of them on price.
Which half should you start with?
Start with the prompts if your team is short on time: the prompts above save hours a week on titles, schema and refreshes without changing what you publish. Start with citations if buyers already ask ChatGPT about your category, which you can check in ten minutes by asking it yourself.
Most B2B teams need both, and the order matters: allow OAI-SearchBot today, because a blocked crawler makes every other step pointless.
Our AI search and SEO service runs both halves for B2B companies. Each month we ask the assistants your buyers use the questions they ask, show you which answers name you, and write the pages those answers draw on. You approve every page before it goes live.
Frequently asked
Can ChatGPT do SEO?
It can do parts of the work: drafting title tags and meta descriptions, grouping a keyword list you paste in, writing schema markup, regex and redirect rules, and outlining a page from a SERP you summarize. It cannot see search volumes, rankings or your Search Console data unless you give them to it, and it can state wrong facts with confidence. Treat it as a fast assistant whose output a person checks.
Is content written with ChatGPT bad for SEO?
Not by itself. Google's guidance from February 2023 says it rewards high-quality content 'however it is produced' and that using AI 'doesn't give content any special gains.' What breaks Google's spam policies is scaled content abuse: generating many pages 'without adding value for users.' Use ChatGPT for drafts and structure, then add your own data, examples and checking.
How do I get my website to show up in ChatGPT search?
OpenAI's help center says that to be eligible, a site must allow OAI-SearchBot to crawl it and its host or CDN must allow traffic from OpenAI's published search bot IP addresses. Placement is not guaranteed: ChatGPT ranks results 'using multiple factors.' After access, write pages that answer buyer questions directly and earn mentions on the sites that answers already cite.
Does ChatGPT search use Google or Bing?
OpenAI says ChatGPT search partners with third-party search providers and rewrites your question into one or more targeted queries it sends to them. Its help page on web search does not name the providers. Being indexed in the major search engines is a sensible baseline, but the only access OpenAI documents as required is letting OAI-SearchBot crawl your site.
Should I block GPTBot?
It depends on whether you want your content used for model training. OpenAI's crawler documentation says the settings are independent: blocking GPTBot tells OpenAI not to use your content to train its foundation models, while allowing OAI-SearchBot keeps your site eligible for ChatGPT search answers. Blocking GPTBot does not remove you from ChatGPT search.
How do I track traffic from ChatGPT?
In your analytics, look for sessions referred by chatgpt.com, and keep a fixed list of the questions your buyers ask so you can run them in ChatGPT each month and log which answers name you. Visits from people who read an answer and then type your URL show up as direct traffic, so referral numbers are a floor, not the full count.



















