GEO Framework

The CITED Framework.

The structured way to earn citations from generative search. Five pillars, organized around the one mechanism that decides what AI systems cite: Query Fan Out.

What is the CITED Framework?

The CITED Framework is a structured approach to Generative Engine Optimization, organized around the Query Fan Out retrieval mechanism. It covers five pillars: Crawl, Inform, Trust, Evaluate, and Distribute. Each one answers a different question about whether AI systems like ChatGPT, Perplexity, Claude, and Google AI Mode will find you, extract you, trust you, measure you, and encounter you.

PillarQuestion it answersHow it maps to Query Fan Out
CrawlCan AI systems find and retrieve your content?Pages must be indexable by Google and Bing, where the sub-queries run.
InformIs your content structured for easy extraction?Structure determines what survives the synthesis step after retrieval.
TrustDo AI systems consider you authoritative enough to cite?Authority determines ranking position in the sub-query results.
EvaluateAre you tracking your AI citations?Monitoring should cover sub-query visibility, not just head terms.
DistributeAre you present where AI systems actually look?Multiple surfaces capture different sub-queries from the same fan-out.

Most GEO advice stops at the first three pillars. Crawl, Inform, and Trust are necessary, but they only decide whether your content deserves to be cited once it is found. Evaluate tells you whether it is working. Distribute determines whether AI systems encounter you across enough sub-queries in the first place. That last pillar is the one most articles skip, and the data says it is the most important one.

How do AI systems decide what to cite?

Through a mechanism called Query Fan Out. AI systems are not search engines. They do not keep their own web index. When ChatGPT, Claude, Perplexity, or Google AI Mode needs to answer a question, the system decomposes the prompt into multiple focused sub-queries, runs each one against Google or Bing, and synthesizes the response from the retrieved results.

This changes which optimizations matter. 95% of fan-out sub-queries have zero traditional search volume, so no keyword tool will show them. And 32.9% of pages cited by ChatGPT appeared only in fan-out sub-query results, not in the original prompt search. The CITED Framework is organized around this mechanism because it is the reason every pillar works.

The five pillars

Pillar 1 of 5 · Crawl

Can AI systems find and retrieve your content?

Crawl is whether AI systems can find and retrieve your pages at all. AI systems run their sub-queries against Google and Bing, so your content has to be indexable in both. If a crawler is blocked, or the page only renders after JavaScript, you are invisible before the rest of the framework even matters.

How it maps to Query Fan Out

Pages must be indexable by Google and Bing, where the sub-queries run.

What to do

  • Allow GPTBot, ClaudeBot, and PerplexityBot in robots.txt.
  • Submit your sitemap to Bing Webmaster Tools. 87% of SearchGPT citations match Bing’s top organic results.
  • Serve content server-side so crawlers see it without executing JavaScript.
  • Add Schema.org markup and define a canonical URL.

Pillar 2 of 5 · Inform

Is your content structured for easy extraction?

Inform is whether your content is structured so an AI system can lift a clean, self-contained answer out of it. Retrieval systems chunk content at heading boundaries, then synthesize a response from the pieces. Structure determines which passages survive that synthesis step and which get dropped.

How it maps to Query Fan Out

Structure determines what survives the synthesis step after retrieval.

What to do

  • Open every section with the answer first, in 40 to 60 words.
  • Keep sections to 75 to 150 words between headings, the optimal range for extraction.
  • Phrase H2 headings as questions so they match the sub-queries AI systems generate.
  • Add FAQ schema, comparison tables, and named entities. Content with 15 or more connected entities shows 4.8x higher selection probability.

Pillar 3 of 5 · Trust

Do AI systems consider you authoritative enough to cite?

Trust is whether AI systems rate you authoritative enough to cite once they have found you. Authority decides ranking position in the sub-query results, and AI systems scan for agreement across independent sources before they cite. A site that contradicts itself across platforms gets excluded.

How it maps to Query Fan Out

Authority determines ranking position in the sub-query results.

What to do

  • Build domain authority and topical depth on the subjects you want to own.
  • Keep content fresh. AI systems cite URLs roughly 400 days newer than organic results. Refresh every 90 to 120 days.
  • Earn cross-source consensus: consistent positioning across your site, G2, Reddit, and LinkedIn.
  • Publish original research, name your frameworks, and attribute authors.

Pillar 4 of 5 · Evaluate

Are you tracking your AI citations?

Evaluate is whether you are measuring what AI systems actually say about you. This is the pillar most businesses are furthest behind on. Between 40% and 60% of cited domains change within a single month for identical queries, so a citation won is not a citation kept. Without tracking, you are optimizing blind.

How it maps to Query Fan Out

Monitoring should cover sub-query visibility, not just head terms.

What to do

  • Track citations across ChatGPT, Perplexity, Google AI Mode, and Claude. Each cites a different mix of sources.
  • Use 30 or more query samples per topic. Single point-in-time checks are unreliable.
  • Watch for citation decay and resurfacing, not just first appearance.
  • Find the queries where competitors get cited and you do not, then close the gap. ILLIXIS tracks AI citations across platforms so you can.

Pillar 5 of 5 · Distribute

Are you present where AI systems actually look?

Distribute is whether you appear across enough surfaces to be retrieved by many different sub-queries. It is the highest-impact pillar and the one most GEO advice skips entirely. 82% of AI citations come from earned media, not your own website. You can optimize your own site perfectly and still barely get cited, because AI systems preferentially cite sources that are not you.

How it maps to Query Fan Out

Multiple surfaces capture different sub-queries from the same fan-out.

What to do

  • Get listed on directories: G2, Capterra, Product Hunt. Each listing is another surface for a sub-query.
  • Participate authentically on Reddit and YouTube, where Perplexity and Google AI Mode look first.
  • Earn third-party mentions and guest posts. They carry more citation weight than anything on your own domain.
  • Keep positioning consistent across every surface so AI systems read agreement, not contradiction.

The deepest move inside Distribute is treating marketing content as primary research: methods-lite papers, ScholarlyArticle schema, Zenodo DOIs, and citation files. ILLIXIS calls this Academic Citation Infrastructure (ACI). For the step-by-step template, read the ACI Playbook, or the peer-reviewable methods paper for the formal framework and experimental design.

CITED Framework FAQ

▶What does CITED stand for?

CITED stands for Crawl, Inform, Trust, Evaluate, Distribute. Each letter is a pillar of Generative Engine Optimization, and together they cover everything that determines whether AI systems find, extract, trust, measure, and encounter your content.

▶Is the CITED Framework the same as SEO?

No. The shared foundations of SEO and GEO are real, but each discipline has its own priorities. SEO has mobile optimization, backlinks, and click-through optimization. GEO has answer-first structure, named entities, schema, and third-party distribution. If your SEO is solid, GEO is a second checklist layered on top, not a replacement.

▶Which pillar matters most?

Distribute. 82% of AI citations come from earned media rather than brand-owned content, and Query Fan Out explains why: more surfaces mean more sub-queries covered. Crawl, Inform, and Trust decide whether your content deserves to be cited once found. Distribute decides whether AI systems encounter you in the first place.

▶What is Query Fan Out?

Query Fan Out is the retrieval mechanism behind AI citation. AI systems do not search your exact prompt. They decompose it into multiple focused sub-queries, run each one against Google or Bing, and synthesize an answer from the results. 32.9% of pages cited by ChatGPT appeared only in fan-out sub-query results, not in the original prompt search. Every pillar of the CITED Framework is organized around this mechanism.

▶Is the CITED Framework a product?

The CITED Framework is an open methodology, not a product. ILLIXIS uses it to organize how content earns AI citations, and the ILLIXIS platform tracks citations across ChatGPT, Perplexity, Google AI Mode, and Claude under the Evaluate pillar. You can apply the framework with any toolset.

Go deeper

The CITED Framework is how ILLIXIS thinks about earning AI citations. Read the argument behind it, then operationalize the highest-leverage slice.

Start with Don’t Just Rank. Get CITED. for the full case, then work the ACI Playbook for the step-by-step Distribute template, or the ACI methods paper for the formal framework and experimental design.

Ready to lose the stack?

One platform. You approve. ILLIXIS executes. Marketing that just happens.

ILLIXIS collects your email to add you to the early-access list and contact you about ILLIXIS. By submitting you acknowledge our Privacy Policy and Notice at Collection.

Join the waitlist•No spam, ever•Unsubscribe anytime
★First 20 founding members: 50% off any plan for your first year.

Marketing, Unstacked.