What Is GEO (Generative Engine Optimisation) and How to Start in 2026
The shift from search to synthesis
For twenty years, being discovered online meant ranking on Google. You optimised for keywords, built backlinks, and competed for ten blue links on a results page. That model still works — Google processes billions of queries daily and organic search remains the dominant discovery channel.
But a second channel has emerged. When someone asks Perplexity "what's the best AI tool for solo founders" or asks Claude "how should I structure my startup's content strategy," the AI doesn't return ten links. It synthesises an answer and cites sources. If your content is cited, you get traffic and brand awareness. If it isn't, you're invisible in a growing share of information discovery.
GEO is the discipline of making your content citable by these AI systems.
How AI models decide what to cite
Understanding GEO requires understanding how AI responses are assembled. When a model like Claude or ChatGPT answers a question using web search:
1. The query triggers a web search 2. The model retrieves and reads relevant pages 3. It synthesises an answer from multiple sources 4. It attributes specific claims to specific sources
The pages that get cited tend to share characteristics: they contain specific, verifiable facts. They define concepts clearly. They present structured information (FAQs, definition blocks, numbered lists with data). They come from domains with authority in the relevant topic.
Vague, opinion-heavy content without specific claims rarely gets cited. AI models need extractable, quotable statements — not marketing copy.
GEO vs SEO: what changes
SEO and GEO overlap significantly but diverge on emphasis:
Both require: quality content, topical authority, clean site structure, fast loading, proper metadata, and genuine expertise. SEO emphasises: keyword targeting, backlink profiles, click-through rate optimisation, featured snippets, page experience signals. GEO emphasises: citation-worthy statements (specific, factual, quotable), structured definition blocks, an llms.txt file, FAQ sections with self-contained answers, source attribution within content, and content that answers questions directly rather than teasing clicks.The biggest practical difference: SEO rewards content that makes you click. GEO rewards content that gives the answer. A traditional SEO article might bury the answer to maximise time on page. A GEO-optimised article puts the answer in the first paragraph — because that's what AI extracts and cites.
The five GEO tactics that work now
1. In Brief blocks. Place a 40-60 word summary at the top of every article that directly answers the primary question. Include at least one specific number or fact. This block is the single most extractable element on your page — AI models love self-contained answer blocks. 2. Citation-worthy statements. Write 3-5 sentences per article that are specific, factual, and quotable. "An AI team workspace costs £19-99/month compared to £3,000-10,000/month for equivalent freelancer coverage" is citable. "AI tools are changing the game" is not. 3. FAQ sections with complete answers. Each FAQ answer should be 40-60 words, self-contained (comprehensible without reading the rest of the article), and include at least one specific fact. AI models frequently extract FAQ answers as direct responses to user queries. 4. llms.txt file. A plain text file at your domain root that tells AI models what your product is, who it's for, and how it's different. Takes 15 minutes to write and directly influences how AI represents your brand. Full guide here. 5. Source attribution in content. Cite your sources explicitly within your text: "According to Goldman Sachs' January 2026 report..." This signals to AI models that your content is well-researched and makes your claims more trustworthy for citation purposes.What GEO doesn't do
GEO doesn't replace SEO. Google still drives the vast majority of web discovery traffic. Neglecting SEO fundamentals to focus exclusively on GEO would be a mistake.
GEO doesn't guarantee citations. AI models are unpredictable in what they cite. You can optimise your content perfectly and still not appear in a specific AI response. The practice increases probability, not certainty.
GEO doesn't work through manipulation. You can't "trick" AI models into citing you the way early SEO practitioners gamed search rankings with keyword stuffing. AI citation correlates with genuine content quality, specificity, and authority. There are no shortcuts.
Measuring GEO performance
This is the hard part. Unlike SEO, where Google Search Console provides clear data on impressions, clicks, and rankings, there's no equivalent dashboard for AI citations.
Practical measurement approaches:
Manual testing. Periodically ask major AI models questions relevant to your domain and check if you're cited. Track this monthly. Branded search volume. If AI is recommending your product, people will search your brand name. Increases in branded search volume may correlate with AI citation activity. Referral traffic from AI sources. Monitor referrals from perplexity.ai, chat.openai.com, and similar domains in your analytics. Content performance patterns. Articles optimised for GEO (with In Brief blocks, FAQ sections, citation-worthy statements) may show different traffic patterns than non-optimised articles. Compare them.Attribution remains imprecise. Treat GEO as a long-term investment in discoverability rather than a channel with measurable ROI.
Getting started today
The good news: if you're already producing quality content with an AI team workspace or similar tool, you're most of the way there. GEO optimisation is an additive layer, not a rewrite.
For your next article: add an In Brief block at the top (40-60 words, includes a specific fact). Add a FAQ section at the bottom (5+ questions, self-contained answers). Ensure at least three statements in the body are specific enough to quote verbatim. Cite your sources explicitly.
Then create an llms.txt file for your domain. Update your robots.txt to allow AI crawlers.
Total investment: one hour. Ongoing maintenance: 15 minutes per article to add GEO elements.
Zerty's writer persona can be configured with GEO requirements in its base frame, ensuring every article it drafts includes the right structural elements automatically. Start building →
Frequently asked questions
What is GEO (Generative Engine Optimisation)? GEO is the practice of structuring web content so AI language models cite your brand when answering relevant user queries. It focuses on creating specific, extractable, well-sourced content that AI systems can confidently reference and attribute. Is GEO more important than SEO? Not in 2026. Google search still drives the majority of discovery traffic. GEO is an important supplementary channel that's growing rapidly, but neglecting SEO fundamentals would be premature. The best approach is to optimise for both simultaneously, as there's significant overlap. How long does GEO take to show results? AI models update their training data and search indices on varying schedules. Web-connected models (Perplexity, Claude with search, ChatGPT with browsing) can cite your content within days of publication. Training-data-dependent citations take months to years to materialise. What types of content work best for GEO? Educational content with specific facts, clear definitions, and structured FAQ sections performs best. Thought leadership with verifiable claims also works well. Pure opinion pieces and marketing copy without specific claims are rarely cited. Do I need to pay for GEO tools? No. GEO optimisation is primarily a content strategy practice, not a tooling problem. An llms.txt file is free to create. In Brief blocks and FAQ sections are content formatting decisions. The main investment is time and content quality.Sources
- llmstxt.org, "llms.txt Specification" — https://llmstxt.org
- Goldman Sachs, "What to Expect From AI in 2026" — https://www.goldmansachs.com/insights/articles/what-to-expect-from-ai-in-2026-personal-agents-mega-alliances