SourceCited / Get cited by ChatGPT
Guide · ChatGPT & Perplexity
ChatGPT cites what it can retrieve and trusts. Get indexed, front-load real answers, and earn corroboration across the sources it reads — then adapt for Perplexity separately.
To get cited by ChatGPT, be retrievable (let OAI-SearchBot/ChatGPT-User/GPTBot in), front-load real, self-contained answers, and be a corroborated source across the independent sites it reads. Perplexity is a separate ecosystem — only ~11% of domains are cited by both — so treat it distinctly, leaning into community sources.
When ChatGPT search answers a live query, it retrieves and cites current web pages — so being citable means being indexed and retrievable, then being a corroborated source the model trusts. Its retrieval leans on a live search index, so pages that rank and are cited elsewhere are the ones it surfaces. Fresh, well-structured content can be discovered within a day or two of publishing.
Get the fundamentals right, then earn corroboration. Concretely: make sure ChatGPT's crawlers (OAI-SearchBot, ChatGPT-User, GPTBot) can fetch you; front-load self-contained answers with real facts; build the same true story across independent sources ChatGPT reads; and keep an entity fact sheet consistent everywhere so the model recognizes you. This is the whole workflow, pointed at one engine.
Act as a buyer researching my category. Answer these questions the way you would for a real user, and after each, tell me: did you name my brand? which sources would you cite? which competitors came up first? Questions: [paste 5 real questions your buyers would ask]. Then list the 5 sources you'd most likely cite for this topic — those are the places I need to appear.
Run the same prompts in ChatGPT, Perplexity, and Gemini — the answers and sources differ per engine.
Perplexity is a separate ecosystem — only around 11% of domains are cited by both ChatGPT and Perplexity, and Perplexity leans harder on community and forum sources. So don't assume one strategy covers both: keep the shared foundation (retrievable, corroborated, consistent entity), then earn presence on the community surfaces Perplexity favors, like relevant subreddits and Q&A threads where your buyers actually ask.
OpenAI runs three distinct user-agents, and mixing them up is the most common self-inflicted wound in this game:
| Bot | What it feeds | Your call |
|---|---|---|
GPTBot | Model training corpora | Your licensing stance — blocking it does not affect citations |
OAI-SearchBot | The ChatGPT search index — the citation pipeline | Allow, always, if you want to be cited |
ChatGPT-User | Live fetches when a user asks about your page | Allow — blocking it breaks direct lookups of your site |
Each has separate rules in robots.txt, so you can refuse training and stay fully citable at the same time.
ChatGPT search skews hard toward recently published or genuinely updated pages — multiple 2026 analyses put the sweet spot inside the last 30 days. That means honest dateModified values plus real refreshes: new data, a new section, an updated example. It does not mean date-stamp gaming, which the March 2026 core update specifically punished. For evergreen money pages, schedule a substantive quarterly refresh and let the date tell the truth.
Make sure its crawlers can fetch you, publish self-contained answers with real facts, and earn mentions across independent sources so the model corroborates you. Fresh, well-structured pages can be picked up within a day or two.
Its live search leans on a web index, so classic SEO — ranking, being indexed, being cited elsewhere — strongly influences what it surfaces. It's not identical to Google, but the fundamentals carry over.
Yes. Only about 11% of domains are cited by both, and Perplexity favors community and forum sources more heavily. Share the foundation, then earn presence on the surfaces each engine prefers.
No. GPTBot governs training data; citations flow through OAI-SearchBot (the search index) and ChatGPT-User (live fetches). The three have separate robots.txt rules, so many sites refuse training and remain fully citable.
There is no fixed timeline. Index inclusion takes days to weeks; freshness effects are immediate but fade; off-page consensus builds over months. And with ~65% of cited sources churning day to day, expect to appear intermittently before you appear reliably — that is the distribution working, not a failure.