Only 19% of New Pages Get Cited by ChatGPT or Claude

Only 19% of new articles earned a citation from ChatGPT or Claude within 60 days, a study of 4,000+ pages found.

The tracker was Searchable, and the write-up comes from an SEO specialist who says the numbers changed how they answer a question clients ask constantly: how long until an AI assistant cites my new page? The honest answer has two halves, and most people only ask about one.

The speed number everyone quotes

Per the study, ChatGPT's median time to first citation was 3.5 days. Of the articles ChatGPT cited within the 60-day window, 26% were cited on day one.

Claude was slower. Its median was 9.5 days, which the author puts at 2.7 times longer than ChatGPT's. When both engines eventually cited the same article, ChatGPT got there first 72% of the time.

Those figures describe the articles that got cited. That qualifier is the part that falls off when the 3.5-day number gets repeated.

The number that changes the forecast

Roughly four in five articles received nothing from either engine inside 60 days. The author calls 19% "the number I keep coming back to," and frames going two months uncited as the most common outcome in the sample — not by itself evidence of a technical problem with the page.

That reframing matters if you are reporting on content performance. A page that has been live for six weeks with no AI citation is behaving like the majority of the sample, not like a failure case. Treating it as a defect sends teams rewriting pages that were never going to clear the bar in that window.

Two questions, not one

The practical takeaway from the piece is to split the question before promising anything. How long a citation takes and how likely your content is to earn one at all are separate problems with separate answers, and only the first has a tidy median attached to it.

For anyone setting expectations with a client or an exec: quote the 19% alongside the 3.5 days, or the timeline number alone will read as a forecast it cannot support.

What is still open

The author flags the next question they want answered — whether articles cited early keep getting cited more often, or whether the early advantage fades. The study as described does not answer that.

Two limits worth holding onto. This is one tracker's sample over one 60-day window, reported by a practitioner rather than published as peer-reviewed work, so treat the percentages as that tracker's measurement rather than an industry constant. And a 60-day window cannot tell you what happens at 90 or 180 days — the four-in-five that got nothing are uncited within the window, not uncitable.

If you are tracking this yourself, the useful instrumentation is a per-URL first-citation date on both engines. Without it you are comparing your results to a median you have no equivalent of.