# GEO vs SEO: 4 takeaways from a debate worth watching > I watched two marketers debate Generative Engine Optimization against traditional SEO. Most of it lined up with what I already believe about AI search — here are the four points that stuck, with the data to back them up. **Published:** Jun 2026 · 7 min read · AI SEO --- I argue about this for a living, so when I came across an episode of The Edward Show where Zachary Long and David Quaid debated Generative Engine Optimization (GEO) against traditional SEO, I watched the whole thing. It's worth your time — and it lines up with a lot of what I've already written here. These are the four takeaways I'd pull out, plus where I'd add some numbers. For two decades the game was simple: find the keywords, earn the links, structure the site, wait for the traffic. LLMs like ChatGPT and Gemini changed how people ask for answers — but, as the debate makes clear, they didn't replace the plumbing underneath. ## 1. AI is a conversational ambassador, not a directory A search engine is a directory. You type "best dentist near me" and get ten optimized options. AI behaves more like a consultant you're chatting with. People don't ask it for a list — they hand it context: "I have severe dental anxiety and I'm terrified of needles, what kind of dentist should I look for?" If your site only optimizes for "Austin dentist," the model has nothing to match that worry against. The point Long makes — and I agree with — is that you're no longer optimizing for a keyword, you're giving the AI the context it needs to recommend you. And this isn't a fringe behaviour: Gartner found that 45% of B2B buyers now use generative AI to gather information on vendors and products. The buyer is describing a problem to a chatbot before they ever type a keyword into Google — the same shift I dug into in [B2B SEO isn't dead, but your TOFU strategy might be](/en/blog/b2b-seo-tofu-strategy). ## 2. The "omniscient AI" is an illusion The most useful myth the two of them puncture: LLMs do not keep a saved copy of the internet. It's computationally impossible to index and store the whole web inside a model's weights. What actually happens is Retrieval-Augmented Generation (RAG): the model strips away the conversational fluff, works out the core query, runs a real-time Google or Bing search in the background, reads the top handful of results, and synthesises an answer. Google even documents the multi-query version of this and gives it a name — "query fan-out". The implication is the whole ballgame: AI is reliant on traditional search to find current information. The data bears it out — Seer Interactive found 87% of SearchGPT citations matched Bing's top organic results, and BrightEdge found 60% of Perplexity citations overlap Google's top 10. So no, SEO isn't dead — it's the infrastructure that feeds AI. If Google can't find your page, ChatGPT can't read it, which is exactly the trap I broke down in [why your B2B company isn't showing up on ChatGPT](/en/blog/why-not-showing-up-on-chatgpt). The full mechanics are in [what AI SEO actually is](/en/blog/what-is-ai-seo). ## 3. The long-tail content loophole for small brands If AI pulls from Google's top results, how does a small shop ever beat a corporation for an AI recommendation? The answer in the debate is the one I'd give too: ultra-specific, genuinely expert content. A local business won't rank page one for a broad head term. But a sharp, first-hand article on something narrow — "how to calm a toddler's toothache at 2am" — can absolutely rank for that specific query. And there's far more of that long tail than people assume: Google has said for years, and recently reaffirmed in the context of AI search, that 15% of the queries it sees every day are ones it has never seen before. When a stressed parent describes that exact situation to an AI, it pulls the article that answers it — and the small brand lands in the response. This is the same "write for the sub-questions an LLM fans out to" idea from [what AI SEO actually is](/en/blog/what-is-ai-seo), and why I keep telling clients that [depth beats broad coverage](/en/blog/b2b-seo-tofu-strategy). ## 4. The fatal trap of "scaled content" This is the point I most wanted to underline. The moment people realise that more content means more surface area for AI to cite, the temptation is to point an LLM at a keyword list and spin up hundreds of articles. Don't. Google calls this scaled content abuse, and its spam policy names it precisely: content where "many pages are generated for the primary purpose of manipulating search rankings and not helping users" — explicitly including "using generative AI tools… to generate many pages without adding value for users." The consequence isn't a gentle nudge: sites that violate it "may rank lower in results or not appear in results at all," up to a manual action. The nuance the debate gets right is that it's not about whether the AI writes "good" sentences — it's about the *means of production*. Mass-manufacturing pages to game the index is the trigger. And if Google drops you, every AI engine that feeds on Google loses you along with it. It's the same reason I'm allergic to [tools that sell you a "score" for content quality](/en/blog/eeat-checker-myth): the thing Google actually rewards is real usefulness, not volume. ## The bottom line The future of organic discovery isn't SEO *or* AI optimization — it's layering them. You need the SEO fundamentals so engines can find and categorise your pages in the first place. Then the content has to be human, context-rich, and genuinely useful enough that when an AI reads it, it trusts you enough to put your name in the answer. GEO sits on top of SEO — [not next to it, and not instead of it](/en/blog/what-is-ai-seo). Worth a watch if you have the time: How Does ChatGPT Cite Brands? Does It Need Google? on The Edward Show.