What Is GEO, and Why Does It Matter Now?

Search is splitting in two. One half still returns a list of blue links. The other returns a written answer, assembled from a handful of sources, with citations underneath. Optimising for the first is SEO. Optimising for the second is what people have started calling GEO — generative engine optimisation.

The distinction matters because the second kind of result has no page two. An assistant either uses your page to build its answer, or it does not.

What actually changes

Most of the SEO foundation carries over unchanged. Fast pages, clean semantic HTML, a sensible heading hierarchy, internal links that describe where they go — all of it still counts, and for the same reasons.

Three things genuinely change.

1. JavaScript is a much bigger liability

Googlebot renders JavaScript, if slowly and with a queue. Most AI crawlers do not. GPTBot, ClaudeBot and PerplexityBot fetch what the server returns and work with that.

This is the single biggest technical difference, and it is why a client-rendered single-page application can rank acceptably in Google while being nearly invisible to an AI assistant. If a fact about your business only exists after a script has run, it does not exist as far as these systems are concerned.

2. Being quotable beats being keyword-dense

A ranking algorithm scores a whole document. A generative engine lifts a sentence or two. That rewards writing that states things plainly:

• Answer the question in the first sentence, then explain.
• Put concrete numbers, prices, timeframes and constraints in text, not in images or diagrams.
• Use question-shaped headings when the content genuinely answers a question.

A frequently asked questions block is the clearest example. It is a format that maps exactly onto how an assistant wants to consume information: a question, followed by a self-contained answer.

3. Structured data describes, rather than decorates

JSON-LD has always been useful for rich results. For generative engines it does something more basic: it tells a machine what kind of thing the page is and how it relates to everything else you publish. An Organization node connected to Service nodes connected to Article nodes is a small, legible map of your business.

What this looks like in practice

None of this requires exotic technology. It requires a site that renders its content on the server, states its facts in text, and describes itself accurately.

That is a fair summary of what a well-built static site does by default — which is why the arrival of AI search has been kinder to plainly built websites than to elaborate ones.

Where to start

If you want a single first move: open your most important page, disable JavaScript in your browser, and reload it. What remains is roughly what an AI crawler sees.

For a lot of sites, that is an uncomfortable experiment. It is also the fastest diagnosis available.