AI search is now responsible for a growing share of how people find information - and almost none of that traffic goes to sites that write for humans reading a full page. Here is what actually determines whether AI cites your content, in language that does not require a technical background.
I had a conversation with a content manager at one of our clients a few months ago. She had noticed something strange in their analytics. Traffic from organic search had stayed flat. But there was a new referral source growing steadily month over month that she had not seen before: chatgpt.com.
She had not done anything differently. No new content, no technical changes. Her existing content was simply structured in a way that AI systems could use.
We spent an hour going through why. What followed was one of the most useful diagnostics I have done in a while - not because of anything complicated, but because of how simple the principles turned out to be once you strip away the terminology.
Here is what I found, and what I think any business needs to understand right now about how to appear in AI search results - and specifically how to get cited by ChatGPT rather than just indexed by Google.
How AI Systems Actually Decide What to Cite
AI search does not rank pages. It selects fragments.
This is the most important thing to understand, and it changes everything about how you should think about content. When ChatGPT, Perplexity, or Google's AI Mode answers a question, it is not picking the top-ranking website and presenting it. It is scanning indexed content, breaking it into pieces, evaluating each piece for relevance and credibility, and assembling an answer from the fragments that pass the test.
Microsoft's principal product manager on the Bing team, Krishna Madhavan, described this in October 2025: AI assistants "break content down, a process called parsing, into smaller, structured pieces that can be evaluated for authority and relevance. Those pieces are then assembled into answers, often drawing from multiple sources to create a single, coherent response."
The consequence is that a well-ranked page written as flowing prose produces fewer usable fragments than a mid-ranked page with clear sections, direct answers, and named sources. The second page is more likely to be cited even though it ranks lower.
Per the Conductor AEO/GEO Benchmarks Report from January 2026, covering 13,770 domains and 17 million AI responses: AI traffic now accounts for 1.08% of all website sessions, growing approximately 1% month over month. That sounds small. Compounding at that rate for twelve months produces a materially different number.
87.4% of that AI referral traffic comes from ChatGPT, per the same report. That concentration matters. If you are going to track one platform first, that is the one.
Why Your Website May Not Be Showing Up in AI Search
There are a few common structural reasons. None of them require a developer to fix.
Your answers are buried. If a heading implies a question and the actual answer is in the third or fourth sentence - after setup, context, or framing - AI systems often miss it or pick a different source that leads with the answer. The fix is to write the direct answer as the first sentence of every section. Everything that follows is supporting detail.
Your content is invisible to crawlers. A significant portion of business websites load key content using JavaScript. AI crawlers, like many SEO crawlers, often cannot execute JavaScript. Content that requires the page to run scripts before appearing may simply not be seen. The test is straightforward: turn off JavaScript in your browser and see how much of your important content disappears.
You are only visible on your own domain. Research from the University of Toronto in September 2025 found that across consumer electronics queries, AI cited third-party authoritative sources 92.1% of the time compared to 54.1% for Google search. AI systems trust independent validation more than brand content. If the only place your expertise lives is your own website, you are at a structural disadvantage in AI search.
Your content is stale. AI systems favour fresh content. A post from three years ago with a 2023 datestamp is a weaker citation candidate than a more recent piece, even with the same quality. Updating high-value existing content - refreshing the date and updating outdated facts - is one of the easier wins available.
Your FAQ content is not structured for extraction. A wall-of-text FAQ section does not help. AI systems look for clean question-and-answer pairs where the answer is immediate and complete. Each FAQ answer should be a single sentence that answers the question fully, followed by optional context. Not the other way around.
What Actually Gets You Cited in ChatGPT and Perplexity
The GEO paper from Princeton, IIT Delhi, and Georgia Tech - published at KDD 2024 and one of the foundational pieces of research in this space, cited widely in subsequent studies including Search Engine Journal's January 2026 AEO deep-dive - tested nine optimisation strategies across generative search systems. The single most effective technique was citing credible sources within the content, which produced a 115.1% improvement in AI visibility for sites that were not already dominant in their category.
That is not an instruction to add a bibliography. It is an instruction to attribute specific claims to named sources with dates. "Organic search drives 1,000% more traffic than organic social" is a weak claim. "Per BrightEdge's 2019 channel report, organic search drives 1,000% more traffic than organic social media" is a citable claim.
The second most reliable factor is structural clarity. Descriptive headings that each cover one specific idea. Sections that can be read in isolation and still make sense. Lists and step formats that give AI clean fragments to extract rather than requiring it to parse a paragraph for the key point.
The research also found what does not work: writing in an authoritative or persuasive tone, using marketing language, or optimising for rhetorical impact. AI systems do not respond to style. They respond to verifiable, well-structured information.
In my digital marketing work, we have been running content audits specifically looking at these factors - whether sections lead with direct answers, whether claims are attributed, whether content is crawler-accessible. The gap between well-structured and poorly-structured content in AI citation rates is not subtle.
How to Set Up Your Content to Get Found in AI Responses
These are the practical steps, in order of impact:
Lead every section with the direct answer. The heading implies a question. The first sentence answers it. Context follows. This one change to existing content produces the most visible improvement in AI citation performance.
Add FAQ schema markup to your key pages. FAQPage structured data tells AI systems explicitly: this is a question, this is the answer. It maps directly to how AI formats responses. If you have a developer, this is a one-hour task for most sites. If you do not, most major CMS platforms have plugins that add it without code.
Check your robots.txt file. AI search engines use specific crawlers with specific names - OAI-SearchBot for ChatGPT search, Bingbot for Microsoft Copilot, PerplexityBot for Perplexity. If any of these are blocked in your robots.txt, your content will not appear in those platforms. Most sites have never configured this deliberately and have no idea what their robots.txt currently allows.
Build presence on third-party sites. Industry publications, trade press, credible directories, and platforms like LinkedIn - content or mentions that appear on domains AI already considers authoritative carry citation weight that your own domain does not yet have. A mention in an industry newsletter with good domain authority is a more useful AI visibility signal than another page on your own site.
Track what is coming from AI. Add utm_source=chatgpt.com to your analytics as a monitored referral. Google Analytics 4 and most analytics platforms will now show ChatGPT referral traffic if you look for it. Bing Webmaster Tools has an AI Performance Report that shows Copilot citation data for free. Knowing where you stand before you make changes is what makes the effort measurable.
The work is not exotic. Most of it is structural. But it requires doing it deliberately, which most content teams have not had a reason to do until now.
If you want the broader context on why this matters alongside your existing SEO work - and what GEO, AEO, and AIO actually mean for strategy - start with GEO, AEO, and AIO: What These Terms Actually Mean and Why Your Business Should Care.
