From SEO to GEO
A Practical Strategy Guide for Writers Navigating the Shift to AI Search
For two decades, being found online meant being indexed and ranked. That assumption no longer holds. A growing share of informational queries are now answered directly by AI systems — ChatGPT, Perplexity, Gemini, Google’s AI Overviews — which synthesise an answer from multiple sources rather than presenting a ranked list of links. Whether or not a piece of writing is read by a human now depends, increasingly, on whether it was first selected, trusted, and cited by a machine.
Two acronyms are worth defining before going further:
SEO — Search Engine Optimisation. The decades-old practice of structuring a website and its content so that traditional search engines, principally Google, rank it highly in their results pages.
GEO — Generative Engine Optimisation. The newer practice of structuring content so that AI systems select it as a trustworthy source when synthesising a direct answer — whether or not the reader ever clicks through to the original page.
Put simply:
SEO optimised for being found. GEO optimises for being remembered.
What actually changes
SEO rewarded volume. GEO rewards authority.
SEO rewarded keyword coverage. GEO rewards conceptual clarity.
SEO rewarded pages. GEO rewards entities.
That last point is the one most writers underestimate. A generative engine isn’t principally trying to rank your page — it’s trying to construct an accurate answer, and deciding, source by source, whether you’re a reliable contributor to that answer. One 2026 analysis found that 28.3% of ChatGPT’s most-cited pages have zero organic visibility in Google, and that fewer than 10% of sources cited across ChatGPT, Gemini, and Copilot rank in the top 10 Google organic results for the same query. Ranking and citation have come apart. A page can win the old game and lose the new one, and vice versa.
Why this favours independent writers, not just large publishers
The old model rewarded whoever could out-produce and out-link everyone else. The new one raises the bar on consistency, verifiability, and depth in a way that favours a focused, identifiable voice over a high-volume content operation — which, properly understood, is good news for exactly the kind of writer reading this.
What I cover in the full paper
The technical baseline — crawlability, and which widely-recommended tactics (llms.txt, schema markup) are genuinely supported by evidence versus oversold
How to structure writing so it survives being chunked and cited by retrieval systems, not just read top to bottom
What it actually takes to become a verifiable author “entity” an AI system can recognise and trust
Platform-specific tensions independent writers face — including a direct look at Substack’s SEO limitations and its “block AI training” toggle
A five-stage practical roadmap, with a monthly checklist
I’ve tried to be honest throughout about where the evidence is solid and where it isn’t — including being upfront that this page exists, in part, because a PDF alone would undercut the paper’s own argument about what AI systems can actually read.