Being findable now means being quotable by a model
Search traffic increasingly arrives pre-answered. What it takes to be the source an assistant cites, not the tenth blue link.
Being quotable by a model means an AI assistant can lift a sentence from your page, use it to answer someone's question, and attribute the answer to you. It is the new shape of search visibility, and it works differently from ranking.
For twenty years the goal was a position: get the page to number one and collect the click. That assumed a person would read the list of results and choose. Increasingly they don't. Google's AI Overviews now appear on a large share of searches, and hundreds of millions of people ask ChatGPT and Perplexity directly. The assistant reads the results, writes the answer, and cites a handful of sources, or none.
From ranking to being cited
Ranking and citation overlap, but they are not the same thing. Most AI Overview citations still come from pages that rank well, so classic SEO is not obsolete. But a meaningful share of citations go to pages outside the top five, which means the model is choosing on more than position. It is choosing on clarity.
The practical consequence: you can rank and still be skipped if your page buries the answer, and you can be cited from further down the results if your answer happens to be the cleanest one available.
The unit of visibility is the passage, not the page
When a model answers a question, it does not quote your whole page. It pulls the specific sentences that answer the question directly and stand on their own. A page written as a slow build toward a conclusion gives a model nothing clean to lift.
This is why passage structure matters more than page length. A long article with the answer in the ninth paragraph is weaker, for citation, than a short section that answers the question in its first sentence.
What makes a passage quotable
Four things, in practice. Answer first: state the answer in the opening sentence of a section, before the context. Stand alone: write so the passage still makes sense when it is lifted out of the page, with no reliance on an earlier 'as mentioned above'. Be specific: a claim with a number or a named example is far more quotable than a general statement. And define plainly: definitions that follow the pattern 'X is...' are easy for a model to reuse verbatim.
None of this is a trick. A model cites the source that most clearly and verifiably answers the question. Being that source is the whole of the work.
The structure a crawler can parse
AI crawlers read your raw HTML, and, importantly, most of them do not run JavaScript. If your content only appears after a script executes in the browser, the model never sees it. Server-render the content that matters.
Beyond that, three things help a model understand who said what. A clean heading hierarchy, so sections map to the questions people ask. Structured data, using Schema.org, that labels your organization, your articles, and your FAQs. And an llms.txt file at the root of your site, giving crawlers a plain-text map of your best pages. We publish all three, and this article is written to the same rules we apply for clients.
The assistants cite differently
There is no single tactic that covers every platform. Google AI Overviews lean heavily on pages that already rank, so traditional SEO plus passage clarity does most of the work there. ChatGPT leans on established, authoritative reference sources. Perplexity leans on community discussion and citations. Only a small fraction of pages get cited by both Google and ChatGPT for the same query, which means AI visibility is won platform by platform, not once.
How to tell if it is working
Ask the assistants the questions your customers actually ask, and see who they cite. Watch your analytics for referral traffic from ChatGPT, Perplexity, and Google's AI surfaces, which increasingly show up as their own sources. And track brand mentions, not just links, because a model can name you in an answer without sending a click.
We restructured The Fragrance Square's product and guide content along these lines and traffic from AI answers rose 31%. The method is repeatable: answer plainly, structure cleanly, and give a model a reason to quote you rather than the next site.
Questions
Why do AI assistants cite some sites and not others?
They cite the source that most clearly and verifiably answers the question, in a passage that stands on its own. Vague or slow-building content gives a model nothing clean to quote.
Does traditional SEO still matter for AI search?
Yes, but it isn't enough on its own. Ranking still helps, but being quoted also needs passage-level clarity, structured data, and answers written up front.
How do I know if AI assistants are citing my content?
Ask the assistants queries in your space and see what they cite, and watch for referral traffic from ChatGPT, Perplexity, and Google AI answers.