--- title: "How to get cited by ChatGPT: what actually works | FWRD" description: "ChatGPT has no submission form and no paid slot. What Google itself requires, and why only 38% of AI Overview citations still come from the organic top 10." url: "https://fwrd.digital/get-cited-by-chatgpt" published: "2026-07-22T11:10:56Z" modified: "2026-07-22T18:28:56Z" --- # How to get cited by ChatGPT: what actually works | FWRD _get cited by chatgpt_ ## There is no submission form for ChatGPT. You cannot buy your way into AI answers. What decides whether you get named: being indexed, being snippet-eligible, and being the clearest answer to one specific sub-question. ## Why does ChatGPT never mention my website? Almost always one of three reasons: your page is not in the index the model consults, it is not allowed to show a snippet, or another page answers the same sub-question more clearly and in fewer words. There is no submission form, no paid placement and no vendor who can put you there. Anyone promising otherwise is selling air. That sounds blunt, but it is good news. What you cannot buy, you can earn. And unlike most SEO folklore, the rules here are written down in public by the companies that generate the answers. ## What does Google actually say about optimising for AI? In May 2026 Google published its own optimisation guide for generative AI features in Search, and it is refreshingly unglamorous. On structured data: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." On special files: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities)." And on cutting your content into bite-sized fragments: "There's no requirement to break your content into tiny pieces for AI to better understand it." What is required sits in the same guide: "To be eligible to be shown in generative AI features on Google Search, a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements." Indexed, snippet-eligible, technically sound. That is the whole list. There is no hidden AEO switch, no mandatory AI file and no schema that pushes you in. ## Why small, focused sites finally have an edge This is the most interesting number of the year. In March 2026 Ahrefs analysed four million AI Overview URLs and found that only 38% of citations now come from the organic top 10, down from roughly 76% in July 2025. The cause is query fan-out: the model splits your question into dozens of sub-questions and searches each one separately. The strategic consequence matters more than the number. If citations are collected per sub-question, the winner is not the site with one strong ranking but the site that covers a whole topic, boring edges included: prices, exceptions, conditions, common mistakes. Breadth across a cluster now beats depth on a single page. Breadth is exactly what a small, focused site can deliver. A freelancer or small business will never out-authority a national portal, but it can be the only page that spells out what something costs, how long it takes and what usually goes wrong. That is a sub-question with one candidate. _What the research says_ - **38%** — of AI Overview citations come from the organic top 10 (Ahrefs, March 2026, 4M URLs. It was around 76% in July 2025.) - **32.8%** — more likely to be cited with clear summaries (Semrush AI Visibility Index 2026, 126M prompts. Correlation, not proof.) - **30-40%** — more visibility from sources, quotes and statistics (GEO experiment, KDD 2024: Princeton, Georgia Tech, IIT Delhi and AI2.) ## What correlates with being cited? Semrush published an AI Visibility Index in 2026 built on 126 million US prompts. Pages that got cited were 22.9% more likely to have strong section structure and 32.8% more likely to include clear summaries and scannable takeaways than pages that did not. Read that carefully: it is correlation, not proof. The study is US-only and observational. Well-structured pages tend to come from teams that work more carefully in general, and that could explain the citations just as well. Treat it as a strong, cheap hint rather than a law of physics. ## Is there one experiment that is actually causal? Exactly one. Researchers from Princeton, Georgia Tech, IIT Delhi and the Allen Institute for AI presented "GEO: Generative Engine Optimization" at KDD 2024. They rewrote the same pages in nine different ways and measured the effect on visibility inside generated answers. Citing sources, adding quotations, adding statistics, writing fluently and using an authoritative voice lifted visibility by roughly 30 to 40%. Keyword stuffing did close to nothing. Two caveats you rarely get handed with the headline: the study used a purpose-built generative engine rather than ChatGPT or AI Mode, and it predates AI Mode entirely. Directional, not conclusive. One finding does travel well though: the biggest gains went to lower-ranked content, which lines up neatly with the Ahrefs number above. ## The silent killer: facts that live only in your JSON-LD This costs more citations than any tactic gains, and almost nobody checks it. On modern sites, prices, opening hours, addresses and ratings often exist only inside structured data: neatly wrapped in a script tag, invisible on the page itself. For Google that is usually fine. For third-party AI fetchers it is not. They pull the raw HTML and routinely strip script tags on the way in, JSON-LD included. What survives is the visible text. Google confirms the other half of the same story in its own guide: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." Structured data locks in nothing you actually need. This one is a log-file observation rather than a vendor-published statistic, which is why we attach no percentage to it. The rule is simple all the same: every fact that matters belongs in visible text. Price in the body copy, not only in Product schema. Opening hours written out, not only in LocalBusiness. Structured data stays useful for rich results, but treat it as a duplicate and never as the only place a fact exists. _Do this_ ## The checklist that changes something. None of these are secrets. They are just rarely done all seven at once. - **Confirm your pages are indexed at all** — Google names indexing as a hard requirement. A single stray noindex switches off your entire AI visibility, and Search Console will show it. - **Never ship nosnippet or max-snippet:0** — The eligibility rule literally says "eligible to be shown in Google Search with a snippet". No snippet means no citation. - **Cover the whole topic, not just the head term** — Query fan-out rewards breadth: one H2 per sub-question, phrased the way a person types it. - **Put every load-bearing fact in visible text** — Prices, conditions, exceptions and opening hours belong in the body, not only in JSON-LD. - **Open every section with a short answer** — Semrush (2026) measured a 32.8% higher likelihood of citation for pages with clear summaries. Correlational, and it costs five minutes. - **Name your sources and your numbers** — The GEO experiment (KDD 2024) measured a 30 to 40% visibility lift from adding sources, quotations and statistics. - **Verify that AI crawlers can actually reach you** — A WAF or bot rule can block them while your robots.txt says allow. Our AI crawlers page explains how to check. ## So where does FWRD fit in? Honestly: FWRD cannot put you inside ChatGPT. Nobody can. What we do is maintain the structure this whole page is about, automatically, which happens to be the part that quietly decays on most sites after three months. An FWRD site is static HTML on an edge network, with a heading per sub-question, a short answer at the top of each section and the facts in visible body copy. The site measures what visitors do, proposes new pages for the sub-questions you do not cover yet, and keeps Search Console and your sitemap current on its own. Indexed and snippet-eligible, the two requirements Google names, become a setting instead of a project. What it is not: a guarantee. We do not sell citations and we do not report an AI visibility score nobody can verify. We make sure the half you control is correct. _Read on_ ## The rest of the AI cluster - [AI SEO explained](/glossary/ai-seo/) — What AI SEO is, and what Google thinks of it. - [Generative Engine Optimization](/glossary/generative-engine-optimization/) — The term behind optimising for AI answers. - [AI crawlers: allow or block](/glossary/ai-crawlers/) — GPTBot, ClaudeBot, PerplexityBot: who trains and who cites. - [llms.txt](/glossary/llms-txt/) — The file where 97% of copies received zero requests. - [SEO at FWRD](/product/seo/) — What is handled automatically. - [The self-evolving website](/product/self-evolving-website/) — How the improvement loop works. ## The structure AI reads, ==maintained for you==. Start free. Your own domain and Google indexing from Grow: 29 euro p/m billed annually.