← Thinking

GEO: optimising for AI search engines is different from optimising for Google

A growing proportion of search behaviour is now happening inside AI interfaces. Users ask ChatGPT a question instead of Googling it. They query Perplexity for a recommendation instead of browsing a results page. They use Google's AI Overviews and get a synthesised answer before they ever click a link.

For marketers and business owners who rely on search traffic, this creates a new problem: the optimisation playbook that works for Google doesn't fully transfer to these generative systems. Not because SEO fundamentals are wrong, but because generative engines prioritise, select, and cite content differently.

Generative Engine Optimisation (GEO) is the discipline of making your content visible, citable, and authoritative in AI-generated search responses. Here's what's actually different — and what you need to do about it.

How generative engines select sources

Traditional Google search uses a link-graph-based authority model: pages rank according to their inbound link profile, technical health, and relevance signals. Generative engines work differently. They select sources to synthesise from based on a combination of:

• Crawlability and clean structure. If a generative engine can't extract clean, well-structured text from your page, it won't use it. Clean semantic HTML, proper heading hierarchy, and logical information flow matter more in GEO than in traditional SEO.
• Factual specificity and citeability. Generative engines favour content that makes clear, specific, verifiable claims. Vague positioning and brand-speak gets filtered out. Specific data points, named processes, and concrete recommendations get surfaced.
• Topic authority signals across the web. If your site is consistently mentioned, referenced, and linked in the context of a specific topic, generative engines are more likely to treat you as an authoritative source on that topic. This is related to but distinct from Google's traditional PageRank.
• Recency and accuracy. Perplexity in particular heavily weights recency. Outdated statistics, deprecated advice, and stale content are less likely to be surfaced in real-time generative searches.

Where GEO and traditional SEO diverge

The keyword paradigm shifts

Traditional SEO optimises pages for specific keyword phrases because Google's algorithm matches search queries to indexed pages. Generative engines don't work this way. They understand intent at the query level and synthesise answers from multiple sources — which means your content doesn't need to "target" a keyword so much as it needs to comprehensively and authoritatively address a topic.

In practice, this means writing content that thoroughly covers a subject from multiple angles — including the questions users typically ask next, the caveats, the comparisons — rather than optimising around a single primary keyword and its variants.

E-E-A-T becomes even more critical

Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) was always important for SEO. For GEO, it becomes the dominant signal. Generative engines are essentially trying to answer: "Who should I trust to tell me about this?" If your site can't demonstrate clear expertise and real-world authority on a topic, it won't be cited.

Practically, this means: named authors with verifiable credentials, About pages that clearly establish expertise, case studies and examples that prove real-world experience, and a consistent publishing history on your core topics.

Structured data and schema markup

Schema markup is more important for GEO than it's ever been for traditional SEO. Generative engines use structured data to understand what a page is about, who wrote it, when it was published, and how it relates to other entities. FAQPage schema, Article schema, and Organisation schema are all worth implementing if you haven't already.

The llms.txt convention

A growing convention among GEO-conscious sites is adding an llms.txt file to the root of the domain — a plain-text document that provides AI crawlers with a clear, structured summary of the site, its purpose, and its key content. Think of it as a robots.txt for language models. It's not universally supported yet, but it's a low-effort signal that's worth including.

What this means for your content strategy

The good news is that the content that performs well in GEO largely overlaps with the content that performs well in traditional SEO: clear, well-structured, authoritative, specific, and genuinely helpful. The difference is in the emphasis.

The sites that will be most visible in the AI-search era are the ones that have built genuine topical authority — not through link schemes or content volume alone, but through being the kind of resource that gets referenced, cited, and recommended by other credible sources in their space.

That's a more demanding standard than traditional SEO. It's also a more defensible one. Authority built on genuine expertise and consistent quality doesn't evaporate when an algorithm changes.

Where to start

If you're new to GEO, the highest-leverage moves are:

1. Audit your site's E-E-A-T signals — author pages, About page, credentials, case studies.
2. Check that your most important pages use clean semantic HTML and appropriate schema markup.
3. Add an llms.txt file to your site root.
4. Identify the three to five topics you most want to be cited on, and make sure you have genuinely comprehensive, specific, up-to-date content on each.
5. Build a plan for external mentions — guest contributions, press coverage, industry directory listings — that establish your brand's presence in the topic graph beyond your own site.

None of this replaces traditional SEO. Google still drives the majority of search traffic for most sites, and the fundamentals of technical SEO, keyword-informed content architecture, and link building remain essential. GEO is an additional layer — one that's becoming increasingly important as search behaviour shifts.