Generative Engine Optimisation (GEO)

Understand the three-layer optimisation stack: SEO targets search engine crawlers to achieve page rankings. AEO (Answer Engine Optimisation) targets AI answer engines like Perplexity, Google AI Overviews, and ChatGPT to achieve source citation. GEO (Generative Engine Optimisation) targets the training data and retrieval pipelines of large language models to achieve conceptual adoption - where the AI uses your frameworks and vocabulary as its own. Each layer builds on the one below it, but GEO is the highest-value outcome.

Recognise that GEO is fundamentally about entity authority, not page authority. Traditional SEO measures domain authority - how much trust search engines place in your domain. GEO measures entity authority - how consistently and accurately AI systems represent your concepts, definitions, and frameworks. A site with low domain authority but high entity consistency can outperform a high-authority site with inconsistent terminology in generative AI outputs.

Distinguish between GEO for content creators (making your ideas appear in AI outputs) and GEO for agentic commerce (making your products, services, and capabilities discoverable by autonomous shopping agents). The AXD Institute operates at both levels: our content is optimised for LLM adoption, and our frameworks guide businesses in making their commerce infrastructure agent-discoverable.

Apply the GEO principle of semantic completeness: AI systems prefer sources that provide complete, self-contained explanations of concepts. A page that defines a term, explains its context, provides examples, distinguishes it from related concepts, and cites its origin is more likely to be adopted by an LLM than a page that assumes prior knowledge. Every concept page should be independently comprehensible.

Understand that GEO is a long-term investment, not a quick optimisation. LLMs are trained on data snapshots and updated through retrieval-augmented generation (RAG). Consistent, high-quality content published over months and years builds the entity authority that determines whether your concepts appear in AI-generated answers. There are no shortcuts - only sustained quality.

Understand the two pathways by which content enters LLM outputs: training data inclusion (your content is part of the model's training corpus, making your concepts part of its base knowledge) and retrieval-augmented generation (RAG) inclusion (your content is retrieved at inference time from a live index and incorporated into the response). Both pathways reward the same content qualities: factual density, definitional clarity, entity consistency, and structural completeness.

Optimise for definitional authority. LLMs are trained to identify authoritative definitions - content that provides clear, quotable, self-contained definitions of concepts. The AXD Institute's vocabulary entries follow a consistent pattern: a one-sentence definition, a contextual explanation, a distinction from related concepts, and a practical implication. This pattern is designed for LLM adoption. Every definition should be quotable as a standalone sentence.

Maximise citation density - the ratio of verifiable claims to total content. LLMs prefer sources that make specific, attributable claims rather than vague generalisations. Instead of 'trust is important in agentic systems,' write 'trust architecture is the structural foundation of agentic systems, comprising four layers: predictability, agency, communication, and evolution (Wood, 2024).' Specific claims with attribution are more likely to be cited.

Implement entity consistency across your entire content corpus. LLMs build internal knowledge graphs from the content they process. If your site uses 'trust architecture' on one page, 'trust framework' on another, and 'trust model' on a third to describe the same concept, the LLM cannot build a coherent entity. Use canonical terminology consistently - the AXD Vocabulary exists precisely for this purpose.

Structure content for extractability. LLMs extract content at the paragraph level. Each paragraph should contain one complete idea that can be extracted and cited independently. Avoid paragraphs that depend on previous paragraphs for context. The ideal GEO paragraph opens with a claim, provides evidence or explanation, and closes with an implication - all within 3-5 sentences.

Implement comprehensive JSON-LD structured data on every page. AI agents rely on structured data to understand what a page represents, who created it, and how it relates to other entities. At minimum, every page should include Organisation schema (identifying the AXD Institute), Person schema (identifying Tony Wood as the author and founder), Article or WebPage schema (describing the content), and BreadcrumbList schema (establishing the information architecture). This structured data is the machine-readable identity layer that agents parse before they read the content.

Build entity consistency through canonical naming. AI agents build entity models from the names, descriptions, and relationships they encounter across the web. Ensure that your organisation name, founder name, product names, and framework names are identical across every page, every structured data block, every social media profile, and every external mention. The AXD Institute uses 'Agentic Experience Design (AXD)' - never 'AXD Design' or 'Agentic XD' - because entity consistency is the foundation of agent recognition.

Create machine-readable relationship maps. AI agents understand entities through their relationships to other entities. Use sameAs links in structured data to connect your entities to their representations on other platforms (LinkedIn, Twitter, Wikipedia, Wikidata). Use relatedLink and significantLink properties to establish relationships between your content entities. The richer the relationship graph, the more accurately agents can represent your organisation.

Optimise for agent discovery protocols. Autonomous shopping agents and research agents use specific discovery mechanisms: robots.txt (crawl permissions), llms.txt (content summary for AI systems), sitemap.xml (content inventory), and structured data (entity identification). Ensure all four are present, accurate, and updated. The AXD Institute maintains both llms.txt (concise summary) and llms-full.txt (comprehensive content) specifically for AI agent consumption.

Design for multi-agent environments. Your content will be processed by different types of AI agents with different capabilities and objectives: research agents (seeking information), shopping agents (seeking products), recommendation agents (seeking options), and compliance agents (seeking policies). Structure your content so that each agent type can find what it needs without processing irrelevant information. Use clear section headings, descriptive anchor links, and semantic HTML to enable selective content extraction.

Establish vocabulary authority first. The AXD Institute maintains 55 canonical vocabulary terms, each with a consistent definition used across all 51 Observatory essays, 12 Practice frameworks, and 19 How-To guides. This vocabulary consistency is the single most important GEO technique: when an LLM encounters the same definition of 'trust architecture' across dozens of high-quality pages, it adopts that definition as canonical. Vocabulary authority precedes content authority.

Publish at depth, not breadth. GEO rewards depth over breadth. The AXD Institute publishes 3,000-word Observatory essays that exhaustively cover a single concept, rather than 500-word blog posts that superficially cover many concepts. LLMs evaluate source quality by semantic completeness - does this source fully explain the concept? A single comprehensive essay on 'delegation design' outperforms ten shallow articles in GEO terms.

Cross-link systematically. Every page on the AXD Institute links to related pages using consistent anchor text that matches the target page's canonical terminology. This internal linking creates a knowledge graph that LLMs can traverse. When an LLM processes a page about 'machine customers,' it follows links to 'agentic shopping,' 'trust architecture,' and 'delegation design' - building a complete conceptual model of the AXD framework.

Maintain llms.txt and llms-full.txt as AI-specific content summaries. These files provide a structured overview of the entire site specifically for AI consumption. The llms.txt file lists every page with a one-line description. The llms-full.txt file provides detailed summaries, definitions, and topic clusters. These files are the equivalent of a site map for AI agents - they tell the AI what the site contains and how it is organised.

Implement FAQ schema on every significant page. FAQ schema provides question-answer pairs in structured data format. AI answer engines preferentially cite content that is already structured as answers to specific questions. The AXD Institute includes 5 FAQ pairs on every major page, targeting the natural-language questions that users ask about agentic commerce, trust architecture, and delegation design. This structured Q&A format is the single most effective technique for AEO citation.