The Three Citation Factors
01 /
Earned Authority
Trusted third-party coverage in publications that AI systems already recognise as credible. 85% of AI-cited links originate from earned media. Without external corroboration, a brand's content is self-assertion, and AI engines systematically deprioritise self-assertion.
02 /
Entity Clarity
The degree to which AI systems can unambiguously identify, categorise, and relate a brand to its category. Brands that AI systems cannot confidently resolve will not be cited confidently, even if they have relevant content and earned media coverage.
03 /
Citation Architecture
The structural design of content so that AI engines can extract and cite specific answers from specific locations on a page. Content that AI crawlers can access but cannot parse or attribute will be ignored regardless of quality.
The RAG Pipeline
Most AI search engines that retrieve live web content use a Retrieval-Augmented Generation (RAG) pipeline. Understanding this pipeline is essential for GEO because it reveals exactly what signals AI engines use to select citation sources.
Query vectorisation
Index search
Candidate filtering
Re-ranking
Response synthesis
Frequently Asked
How does ChatGPT decide what to recommend?
ChatGPT and other AI search engines use a Retrieval-Augmented Generation (RAG) pipeline. When a user asks a question, the system converts the query into a vector embedding, searches its index for semantically relevant content, filters and re-ranks candidates based on quality signals, and then synthesises a response with attributed citations. The selection is not random, it is based on specific, measurable signals including earned authority, entity clarity, and citation architecture.
Why does my brand not appear in AI recommendations?
The three most common reasons a brand is invisible in AI recommendations are: (1) No entity clarity, AI engines cannot confidently resolve who your brand is because you lack consistent structured data across platforms. (2) No citation architecture, your content is not structured for passage-level extraction, so AI engines cannot pull specific answers from your pages. (3) No earned authority, you lack third-party validation from publications that AI engines already trust. 85% of AI-cited links originate from earned media, not owned blog content.
What is the difference between how Google and ChatGPT rank content?
Google ranks pages based on relevance signals (keywords, backlinks, user engagement) and returns a list of blue links. ChatGPT and other AI engines cite sources based on entity authority, content extractability, and third-party validation, and return a synthesised answer with attributed citations. 88% of Google AI Mode citations come from outside the organic top 10, meaning the two systems operate on almost entirely separate selection criteria.
What is a knowledge graph and how does it affect AI recommendations?
A knowledge graph is a structured database of entities and their relationships. Google's Knowledge Graph, Wikidata, and similar systems are used by AI engines to resolve entity identity, to confirm that 'Zennith Agency' is a specific organisation with specific attributes, not just a string of text. Brands that exist as named, resolvable entities in the knowledge graph are cited more confidently by AI engines than brands that only exist as website content.
Apply This to Your Brand