Generative Engine Optimization and SEO: What Changes for Visibility

The term became widely known through a 2023 research paper by a team including researchers from Princeton University, titled “GEO: Generative Engine Optimization”. The authors tested how different content changes affected visibility in generative answers. Adding citations to credible sources, including quotations and adding relevant statistics improved visibility in their experiments, while keyword stuffing did not help.

Since then, GEO has become an umbrella term in the industry, often used alongside answer engine optimization (AEO) and AI optimization (AIO). Definitions vary, but the shared idea is the same: AI answers are a new surface, and content can be shaped to be useful to them.

More than the hype suggests. Most generative search systems retrieve information from a web index before writing an answer, a process often called retrieval-augmented generation. Google’s AI features use its search index and ranking systems. Microsoft Copilot draws on Bing. ChatGPT search and Perplexity use their own crawlers and search partners.

That means the basics still decide whether you are even considered:

• Pages must be crawlable and indexable, with no accidental robots.txt blocks for the crawlers you want.
• Content must be relevant to the query and genuinely helpful.
• Sites with clear topical focus, visible expertise and good reputations are trusted more.
• Technical health matters: fast loading, mobile-friendly pages and clean HTML.
• Structured data helps machines understand entities such as products, organisations, events and FAQs.

A site with poor SEO rarely has good GEO.

The unit of competition shrinks

Classic SEO competes at page level for a ranking position. Generative answers often quote or paraphrase a passage. A page might be strong overall but lose citations because no single section answers the sub-question clearly.

Queries become conversations

People ask longer, more specific questions in AI tools and follow up. Systems such as Google’s AI Mode run many related searches in the background to build an answer. Content that covers the related questions around a topic has more chances to be retrieved.

Brand mentions matter beyond links

Assistants form a picture of which brands are relevant for a topic from many sources: reviews, comparisons, forums, news and directories. Being mentioned in credible third-party content can influence whether a brand is named in an answer, even without a link.

Clicks become less predictable

When an answer satisfies the question, fewer people click. When they do click, they often arrive with clearer intent. Success metrics need to account for both effects.

The practical guidance overlaps heavily with good editorial practice:

• Put a direct answer to the question in the first two or three sentences under each heading.
• Use headings that mirror real questions people ask.
• Include specific, verifiable facts: names, steps, specifications, prices or ranges with context.
• Cite credible sources for claims, as the GEO research found citations help.
• Show first-hand experience: original photos, test results, process details and named authors.
• Keep pages up to date and show the update date.
• Write clear definitions and comparisons, which answers often reuse.

Avoid tactics that try to trick models, such as hidden text aimed at AI crawlers. These risk spam penalties and can damage trust when discovered.

A few technical choices are specific to AI systems. Robots.txt can allow or block individual AI crawlers such as GPTBot, OAI-SearchBot, PerplexityBot and ClaudeBot, and Google-Extended controls use of content for Gemini model training and grounding outside Search. Blocking a search-focused crawler may remove you from that assistant’s answers, so decide crawler by crawler with the trade-offs in mind.

Server-side rendering helps, because some AI crawlers do not execute JavaScript fully. Content loaded only by client-side scripts may be invisible to them. Some sites add an llms.txt file, a proposed standard for summarising a site for language models, though major search providers have not confirmed using it.

Measurement is still immature, so a mixed approach works best.

• Build a fixed prompt set of 50 to 200 questions your customers ask, across awareness, comparison and purchase stages.
• Check them monthly in the assistants that matter to your market, and record whether your brand is mentioned, whether your pages are cited, and which competitors appear.
• Track referral traffic from AI platforms in GA4; chatgpt.com, perplexity.ai and copilot.microsoft.com often appear as referrers.
• Continue tracking Search Console impressions, clicks and positions, since AI Overview traffic is included there.
• Monitor branded search volume as an indirect sign of AI visibility.

Tools from Semrush, Ahrefs and specialist vendors offer AI visibility tracking. Their data is useful for trends, but answers vary by user and moment, so treat figures as samples.