Understanding generative engine optimization
Editor’s note: Arthur Mstoyan is head of SEO at Storyblok.
For years, the gold standard in search engine optimization was a top-ranking spot on a list. But the landscape has changed, and with it, the rules of online visibility. According to research from WordStream, AI Overviews now show up for almost 55% of Google searches and a Semrush study projects that AI search visitors could surpass traditional search visitors by early 2028. This new reality means that visibility is no longer about climbing a list; it’s about being selected, cited and trusted within those generative AI answers.
This requires a new approach: Generative engine optimization (GEO). This is the practical strategy for brands to earn their place by writing for people, expressing meaning for machines and proving their authority.
From rankings to relevance
GEO starts with a simple observation: users increasingly read the summary. If the summary includes your brand accurately, you’re present in the moment that matters. Vague language gives models nothing to hold – specifics do. A company that repeatedly says “our solution” becomes hard to place; a company that consistently describes “custom hats,” “custom lanyards” and “event merchandise” anchors itself in topics machines recognise. The same applies to problems and outcomes. Spell out use cases, not slogans. Consistency across your site and profiles helps engines confirm that the source they’re citing is stable and credible.
Authority becomes less about posture and more about proof. Depth across a topic – definitions, implementation notes, comparisons and evidence – signals expertise far better than isolated thought pieces. Internal linking should reveal how ideas relate, not just pass authority around. The goal is a body of work that explains a subject end-to-end and can be navigated logically by readers and parsers alike.
Technical aspects still matter
- Pages should be fast, crawlable and indexable, with canonical URLs declared where needed (rel="canonical").
- Apply schema that actually reflects the page (e.g., article, how to, product, FAQ, organization, person) and include author, date published and date modified.
- Keep author profiles real and consistent.
- Make claims verifiable and cite sources.
When engines assemble answers, content that is clear, sourced and easy to attribute, it is more eligible to be quoted.
Off-site signals carry new weight
Mentions in reputable publications, appearances in industry roundups, consistent naming across company listings – these aren’t vanity metrics; they are corroboration. When models look for a safe source to cite, brands that are referenced beyond their own domains are easier to select.
Writing for a hybrid audience
GEO doesn’t endorse robotic prose. People still decide whether to trust and act. The work is to write plainly and structure carefully so machines can extract without distorting the intended message.
Lead with the answer, then earn it. Open with a clear definition or conclusion before you widen into context, examples and edge cases. Keep paragraphs tight, offer one idea at a time and cut filler. When you use numbers, name the source. Where you make a claim, show the pathway: What was measured, what changed and what remained the same.
Headings should behave like signposts, not slogans. A restrained H1/H2/H3 hierarchy beats a cascade of subheads. Label sections for what they do: What it is, how it works, when to use it. Let the copy carry the persuasion. Summaries help with extraction, but they should distil, not duplicate. A short recap at the end can reinforce the key idea in language that’s quotable without being trite.
Clusters matter more than ever. Choose the domains where you desire to be heard and cover them thoroughly. A concept piece should point to a practical guide; the guide should point to an implementation note or case study; the case study should link to a comparison that sets boundaries. That network gives readers a route through your thinking and gives engines a graph of meaning to follow.
The final step is earning outside verification. Publish where peers read. Contribute data or methods that others will cite. Keep profiles aligned – company, product, people – so references resolve to you, not a near-match with a similar name. GEO rewards brands that are not only internally consistent but also externally confirmed.
Measuring success in a new era
Success will be measured less by rank and more by presence. While rankings and organic sessions won’t disappear, they won’t tell you whether you were named in the answer a user saw. The more relevant question becomes: For the topics that matter to us, how often are we being surfaced and cited in generative results? This metric will push the industry in the right direction: toward clarity, depth, stable identity and verifiable expertise.
This new metric isn’t just theoretical. Studies show that when an AI Overview is present, the average click-through rate for the top organic link can drop by as much as 34.5%, with some high-traffic keywords seeing a traffic decline of over 60%. As users become more accustomed to zero-click answers, the ability to appear in that initial summary is paramount. Arc Intermedia found that the users who do click through from generative AI results are often more engaged, with early data suggesting they stay on site 8% longer and are 23% less likely to bounce immediately.
We must continue to write for people but express knowledge in a way that machines can interpret without guesswork. Be specific. Be consistent. Use the right markup to expose meaning. Build content clusters that truly explain a subject rather than just decorate it. Aim beyond simply being found; aim to be referenced. In a world where summaries come first, the reference is the result.