A growing share of the questions your customers ask never reach your website as a click. They’re answered inside an AI Overview, ChatGPT, Perplexity or Gemini — which read the web, synthesise an answer, and cite a handful of sources. Generative Engine Optimization (GEO) is the practice of being one of those cited sources. The good news for developers is that most of GEO isn’t a new trick to learn; it’s engineering fundamentals, done properly.
GEO “is still SEO” — and that’s the point
The most useful thing to understand about GEO is what the major search providers keep repeating: optimising for AI answers is, in the main, still SEO. The same things that earn a classic ranking — pages a crawler can reach and parse, clear structure, genuine expertise, trustworthy signals — are what make your content extractable and citable by a language model. There is no llms.txt shortcut that substitutes for that. So the developer’s job isn’t to chase a separate GEO checklist; it’s to remove every technical reason an answer engine might skip your content.
The parts that are squarely a developer’s job
Make content reachable without JavaScript. An answer engine’s crawler may not execute your client-side JavaScript, or may not wait for it. If your core content only appears after a framework hydrates in the browser, you’ve made it invisible to the systems deciding what to cite. This is exactly where a static-first build wins: the content is in the HTML, server-side, from the first byte.
Use real semantic HTML. Headings that describe the actual structure of a page, lists that are lists, a single meaningful <h1>, articles and sections marked up as such — this is how a model understands which passage answers which question. The same semantic markup that makes a page accessible makes it legible to an answer engine. Accessibility and GEO are the same work.
Add structured data. Schema.org markup — usually JSON-LD — explicitly labels what a page’s content is: an FAQ, an article, a product, an organisation. It’s one of the most reliable ways to help both rich search results and AI answers understand and quote your content correctly. Our own FAQ page emits FAQPage structured data for exactly this reason, and this article ships as BlogPosting.
Answer the question early. Language models lift complete, self-contained answers. A page that states its answer plainly in the first paragraph — before the storytelling, before the context — is far easier to cite than one that buries the point halfway down. Write the summary first; earn the reader’s attention for the detail second.
What GEO is not
It isn’t keyword stuffing wearing a new hat, and it isn’t a plugin you install. If an “AI optimisation” tactic depends on tricking the model rather than genuinely being the best available source, it’s the kind of shortcut providers have explicitly warned against — and the kind that ages badly. Being cited by an answer engine is earned the way search visibility has always been earned: useful, well-structured, trustworthy content, served fast and marked up honestly.
That’s convenient, because it’s what good engineering produces anyway. When we build a site, static-first delivery, semantic HTML, structured data and answer-first content aren’t a GEO add-on — they’re the default. And when the goal is actively growing that AI-search visibility over time, our sister agency Ripple18 runs it as an ongoing discipline. Built by Wave18, scaled by Ripple18.