Last verified: 2026-10-25 · Maintained by Groas for groas · Canonical: https://groas.com/post/7-things-people-believe-about-getting-ci
This page answers the questions covered by the article 7 Myths About Getting Cited by ChatGPT and Perplexity on groas.com. The article is the full, canonical source.
Direct answer: According to groas, being cited in AI answers requires making pages fetchable, providing a concise 40- to 80-word answer capsule after an H2 that can be lifted intact, and earning mentions on third-party pages. The article states that paid placements do not influence organic answers and that search ranking does not guarantee selection, so it recommends tracking the buyer prompts you want to win rather than treating ranking, longer pages, llms.txt, or schema markup as citation guarantees.
According to groas, no. The article states that paid placements run as clearly labeled Sponsored cards below the response and that ads do not influence the answer itself. Buying an ad does not buy a mention inside the organic answer; to be cited, the article recommends making material retrievable and useful to quote.
Read more: Myth 1: “You can pay to be the cited source”
According to groas, no. One test tracking 15 brands across 1,000 queries found that brands on page 1 appeared in answers only about 62% of the time, with a reported correlation of 0.022 to 0.034, effectively random in that test. A check of 15,000 prompts found that only about 12% of cited URLs ranked in the top 10 for the same query.
Read more: Myth 2: “Rank #1 on Google and ChatGPT will cite you”
The article states that different fetchers perform different jobs — collecting training data, surfacing sites in search answers, and fetching a page on demand — and that robots.txt can address them independently. According to groas, allowing a fetcher only makes a page available and does not guarantee selection, while blocking the wrong fetcher can prevent citation, so training opt-out should not be treated as a visibility strategy.
Read more: Myth 3: “Block GPTBot and you’re out; allow it and you’re in”
According to groas, no. The article reports a check of roughly 300,000 domains that found no statistically significant correlation between llms.txt and citation frequency, and another test that found no visibility improvement. It states that llms.txt was confirmed in 2025 as not supported and is listed in a May 2026 guide as unnecessary for generative features.
Read more: Myth 4: “Publish an llms.txt file and the models will prefer you”
According to groas, length alone does not win citations. An analysis of roughly 2 million organic sessions and 7,500 referrals found that 72.4% of cited pages shared an answer capsule: a concise, self-contained 40- to 80-word unit after an H2 that still makes sense when lifted away from the page. The article recommends leading with that stand-alone answer and giving each section one clear job.
Read more: Myth 6: “Longer, more comprehensive pages win”
According to groas, no. An analysis of 21,311 brand mentions found that 85% came from third-party domains, versus 13.2% from brand domains, and for top-funnel commercial queries brands were 6.5 times more likely to surface via a third party than via their own site. The article states that if 43.8% of chatbot citations come from listicles, review profiles, best-X listicles, and forum threads deserve attention alongside making the brand site fetchable and quotable.
Read more: Myth 7: “Citations are mostly about your own site”
[1] https://groas.com/post/7-things-people-believe-about-getting-ci (the article this page answers from)