Yes — ChatGPT cites websites, but only some of the time. When ChatGPT browses the live web to answer a query, it links the sources it used. When it answers from knowledge learned during training, there are usually no citations because no specific page was consulted.
Understanding when citations appear — and what makes a page citable — is the key to earning them.
When ChatGPT cites sources
ChatGPT draws on two channels:
- Live browsing/retrieval. For current, factual, or specific queries, ChatGPT searches the web, grounds its answer in real pages, and cites them. This is when website citations appear.
- Trained knowledge. For general questions it can answer from memory, ChatGPT responds without browsing — and therefore without citations, even though that knowledge originally came from web content.
So the practical question isn’t “does ChatGPT cite?” but “is this the kind of query that triggers browsing, and is my page good enough to be the cited source?”
What makes a page citable
When ChatGPT does browse, it favors pages that are:
- Directly answering the query. The answer is stated clearly and early, not buried.
- Authoritative. From a source with genuine E-E-A-T and reputation.
- Specific and attributable. Concrete facts, data, and claims it can quote and attribute.
- Retrievable. Crawlable, fast, and not hidden behind scripts or paywalls.
See writing for AI citation for how to structure content this way.
How to become a cited source
- Answer real questions directly. Lead with the answer; elaborate after.
- Publish original, attributable facts. A specific, sourceable statistic is the most citable unit of content.
- Build authority. Earn reputable mentions and links so ChatGPT trusts you enough to cite.
- Stay fresh and crawlable. Keep cornerstone pages current and accessible.
- Track it. Monitor whether ChatGPT cites you for your priority queries — see optimizing for ChatGPT.
The crawler that decides whether you’re eligible
Being citable in ChatGPT is not the same permission as being usable for training, and OpenAI keeps them apart. Its crawler documentation describes OAI-SearchBot as the agent that builds the index ChatGPT’s search feature retrieves from — distinct from GPTBot, which collects content for model training, and from ChatGPT-User, which fetches a page because someone’s question triggered it right then.
The practical consequence catches people out: if your robots.txt disallows OAI-SearchBot, you are not in the index ChatGPT searches, and no amount of content quality will make you a cited source. Meanwhile disallowing GPTBot alone costs you training influence but leaves citation eligibility intact. Check which of the three your rules actually name before concluding that ChatGPT “won’t cite you.”
Two kinds of “citation”, and why the distinction matters when you measure
When you start tracking ChatGPT citations, you will meet a measurement trap. There are two very different things a monitoring tool can call a citation:
- A real cited URL. The engine browsed, retrieved specific pages, and returned links. This is a fact about what the engine did.
- A domain named in prose. The engine answered from trained knowledge and simply mentioned your brand or site in the text. No page was fetched; nothing was linked.
Both feel like a win in a screenshot, and only the first one is evidence of retrieval. LLM Metrix records the difference rather than flattening it: when a scan runs against a search-grounded engine, Citation Intelligence stores the actual URL the engine returned; when the answer came from an ungrounded call, it records the domain the analyzer extracted from the answer text and there is no URL to store. A report that shows you an exact page path is telling you something stronger than one that shows a bare domain, and you should read them differently.
This matters most when you compare engines. Perplexity cites on nearly every answer; ChatGPT cites only on the subset of queries that trigger browsing. If your tracking counts prose mentions and real links as one number, ChatGPT will look far more citation-friendly than it is, and you will optimize against a metric that isn’t measuring retrieval at all.
Frequently Asked Questions
Does ChatGPT show its sources?
ChatGPT shows source citations when it browses the web to answer a query. When it answers from trained knowledge without browsing, it typically does not show specific citations.
Why doesn’t ChatGPT cite a source for some answers?
Because for many general questions it answers from knowledge learned during training rather than consulting a specific page. With no page retrieved, there’s no source to cite.
How do I get ChatGPT to cite my website?
Publish authoritative content that answers the query directly, includes specific attributable facts, and is crawlable and fresh. Citations appear when ChatGPT browses, so being the clearest, most trustworthy retrievable source for a query is what earns them.
Does being cited by ChatGPT drive traffic?
It can, when users click the citation, but the larger value is visibility and credibility — your brand appearing as a trusted source inside the answer, even when there’s no click.