What GEO actually is, and what it borrows from SEO

How an assistant picks a source

Ask ChatGPT or Perplexity a commercial question and three things happen in sequence. The model rewrites your question into one or more search queries. A retrieval layer pulls candidate documents. The model then drafts an answer from passages inside those documents and attaches links to the passages it leaned on.

Every one of those steps is a filter. If a crawler cannot fetch you, you never enter the candidate pool. If your answer is spread across six paragraphs, the passage that gets retrieved carries half the meaning. If the claim has no source, the model prefers the competitor who cited one.

What carries over from SEO

Crawlability, sitemaps, clean HTML, fast rendering, internal links, unique titles. None of that stopped mattering. Retrieval layers are still built on indexes, and several of them are literally Bing and Google.

The practical implication: if a site scores badly on classic technical SEO, GEO work is premature. Fix access first.

What is genuinely new

Passage self-containment. A paragraph that starts with 'as we mentioned above' is unusable when lifted out of the page. Write each section as if it will be read alone, because it will be.

Evidence density. Models are tuned to prefer text that looks verifiable — named authors, dated figures, linked sources. 'Industry-leading uptime' loses to '99.95% uptime measured across 2025, status page linked'.

Off-site consensus. Assistants quote Reddit, Stack Exchange, review sites and trade press constantly. A brand with no third-party footprint is hard to describe, so it gets described less.

How to start this week

Name the AI crawlers in robots.txt. Add Organization and FAQPage JSON-LD. Rewrite the opening of your five highest-intent pages so the answer sits in the first two sentences. Then measure: run the same five category prompts against an assistant once a month and record who gets named.