By the EPR Editorial Team. Updated October 12, 2026.
To rank on ChatGPT, make sure OpenAI's search crawler can read your site, state your facts clearly, and earn coverage in the outlets ChatGPT already cites for your category. OpenAI says any public website can appear in ChatGPT search and has not published a ranking formula, so the steps below combine OpenAI's documented guidance with what EPR's citation research shows. Track results by running a fixed prompt set every two weeks.
For how OpenAI's engine picks the brands it names, see Inside ChatGPT's Brand Bias. For the domains ChatGPT cites most, see The ChatGPT Citation Source Index 2026. If a competitor is named and you are not, start with Why ChatGPT Recommends Your Competitor: A 5-Test Diagnostic.
Key Takeaways
- Allow OAI-SearchBot. Sites that opt out of it will not be shown in ChatGPT search answers.
- GPTBot is a separate choice. It covers model training, and OpenAI says each setting is independent.
- Independent sources decide who is named. Wikipedia, major publishers and Reddit show up often in EPR's citation research.
- Visibility differs by engine. Test the same prompts on ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews.
- Measure every two weeks with a frozen prompt set, and track referrals in analytics.
How does ChatGPT choose which brands to name?
ChatGPT builds answers from two sources: what the model learned in training and what it retrieves from the web when search is used. OpenAI documents separate crawlers for these jobs, and the settings are independent.
| OpenAI user agent | What it does | Effect of blocking it |
|---|
| OAI-SearchBot | Surfaces sites in ChatGPT search | Site is not shown in ChatGPT search answers |
| GPTBot | Collects content that may be used for model training | Opts the site out of training use, with no effect on search |
| ChatGPT-User | Fetches a page when a user action triggers it | Stops user-triggered page fetches |
OpenAI also says it may surface a link and title for a blocked page if it learns the URL from a third-party search provider. Which brands get named among eligible pages is not documented. EPR's citation indexes show three recurring patterns: Wikipedia and other structured reference sources, publishers that have content deals with OpenAI, and Reddit on opinion and "best of" questions. Treat those as strong tendencies from EPR's testing, not guarantees from OpenAI.
What changed on ChatGPT in 2026?
Two changes matter for anyone planning a visibility program.
- Models turned over. OpenAI's help center says GPT-4o, GPT-4.1, GPT-4.1 mini, o4-mini and GPT-5 were retired from ChatGPT on February 13, 2026. Build tests around prompts and sources, not model names.
- Custom GPTs are being replaced by plugins. OpenAI plans to retire custom GPTs, so a branded GPT is no longer a safe distribution plan. See ChatGPT Plugins vs Custom GPTs for PR Agencies.
The eight steps to get cited
1. Let OAI-SearchBot crawl your site
Check robots.txt before anything else. A blanket rule that blocks every AI bot removes you from ChatGPT search answers, even if you only meant to stop training. OpenAI recommends allowing OAI-SearchBot and its published IP ranges. Blocking GPTBot does not remove you from search, because the settings are independent. See OpenAI's crawler overview and EPR's guide to robots.txt as a business decision.
2. Make your entity facts consistent
Ask ChatGPT "What is [brand]?" and fix what is wrong at the source. Your site, leadership profiles, LinkedIn, Crunchbase and press mentions should state the same category, founding date, location and leaders. A well-sourced Wikipedia article helps where you meet notability rules, which require independent coverage, not press releases. Build the press first, then the article.
3. Earn coverage in the outlets ChatGPT cites for your category
List the domains ChatGPT cites when it answers your buyers' questions, then pitch those outlets first. OpenAI has announced content partnerships with publishers including the Associated Press, News Corp, the Financial Times, Axel Springer, Vox Media, Condé Nast, Hearst, Time, Le Monde, Prisa Media and Dotdash Meredith, and EPR's research finds those outlets over-represented in answers. Confirm the current partner list before building a pitch plan, since deals change.
4. Publish answer-first pages on your own domain
Own the definitional questions about your brand. An About page that states in its first sentence what the company is, who it serves and where it operates gives retrieval something clean to quote. Put the answer in the first 40 to 80 words under each heading, and support it with named sources.
5. Add structured data
Use Organization, Person, Article, FAQPage and BreadcrumbList markup, with sameAs links to your Wikipedia, Wikidata, LinkedIn and Crunchbase profiles. Structured data helps parsers read who you are. It does not guarantee a citation.
6. Publish original research
Engines can quote data that exists nowhere else. A benchmark, an index or a survey with a clear method gives publishers a reason to cite you, and their coverage then supports step 3. One annual study usually does more than a dozen opinion posts.
7. Use named authors with real credentials
Attribute content to people with verifiable expertise, and mark them up with Person schema linked to their professional profiles. Anonymous pages give a retrieval system less to trust.
8. Take part in communities honestly
Reddit carries weight on recommendation and comparison questions. Real staff and founders answering questions in the open build a record. Astroturfing gets detected and banned, and it leaves a trail that works against you.
What does not work?
- Keyword stuffing. Retrieval reads context, not density.
- Press-release syndication on its own. Wire copy rarely changes who gets named. Independent coverage does.
- Hidden instructions in page text. Attempts to tell ChatGPT what to say are poor practice and put your credibility at risk.
- Asking ChatGPT to remember your brand. Memory is per user and does not change the model.
- Submission services. OpenAI's publisher guidance describes no submission process beyond not blocking its crawler. Anyone selling guaranteed inclusion is selling a promise OpenAI does not make.
How long does it take?
These timelines are EPR estimates from client and index work, not OpenAI figures.
- Robots.txt changes: about a day to register, according to third-party guidance on OpenAI's crawlers.
- Schema and on-page changes: weeks.
- New pages entering search answers: roughly two to three months.
- Wikipedia article: two to six months from sourcing to pickup.
- Tier-1 coverage: three to six months before a measurable change.
- Training-data effects: depend on OpenAI's release cycle, and you cannot schedule them.
How do you measure progress?
Build a frozen set of 15 to 50 prompts that mirror real buyer questions, and run it every two weeks across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews. Log whether the brand is named, how many of the named brands are yours, and which sources are cited. Report monthly and reallocate quarterly. This is the method behind the Citation Share Index and the AI Platform Citation Source Index 2026, and Share of Model is the metric that replaces share of voice.
In analytics, OpenAI says ChatGPT appends utm_source=chatgpt.com to outbound links, so referral traffic can be isolated in Google Analytics. See OpenAI's publisher FAQ.
The ChatGPT checklist
- OAI-SearchBot allowed in robots.txt, and a deliberate decision made on GPTBot.
- Entity facts consistent across your site, profiles and press.
- Tier-1 coverage in the outlets ChatGPT cites for your category within the last twelve months.
- A definitional About page that answers "what is [brand]" first.
- Organization, Person, Article and FAQPage schema deployed.
- One original research asset with a published method.
- Named authors with Person schema.
- Real staff presence in the communities your buyers use.
- A frozen prompt set run every two weeks.
- Referral traffic from utm_source=chatgpt.com tracked in analytics.
The Five-Engine Cluster
This playbook covers how to get cited by ChatGPT. For how communications teams use it day to day, see ChatGPT for Public Relations: The Complete 2026 Guide, How PR Teams Use Claude, How PR Teams Use Perplexity and How PR Teams Use Copilot.
