Visible when the AI answers.
More and more purchase decisions start with a question to ChatGPT, Perplexity or Google AI Overviews. The sources that get cited are machine-readable, fact-based and recognizable as an entity. We build exactly that — in code.
What is Generative Engine Optimization?
SEO optimizes for a list of results. GEO optimizes for an answer — and for which sources get named in it.
Generative Engine Optimization (GEO) covers every measure that makes a website appear as a source in AI-generated answers — in ChatGPT, Perplexity, Google AI Overviews or Copilot. Unlike classic SEO, GEO does not aim at rankings but at citation: the AI has to retrieve the content, understand it, rate it as trustworthy and name it in its answer. That is decided by bot access, structure, facts and entity signals.
For shop owners this means: questions like "Which Shopware agency fits a B2B merchant?" or "Which shop ships product X to Austria?" are increasingly answered without a click on a results list. Whoever is missing there does not exist for that customer — regardless of their Google ranking.
How do shops get cited in AI answers?
AI systems cite pages they are allowed to retrieve, understand unambiguously and can classify as a credible entity. Concretely: open bot access, structured data, fact-based direct answers and consistent company signals outside your own website. Which source an engine prefers differs per system — which is why we measure citations per engine instead of just Google rankings.
Citable facts
AI systems prefer concrete numbers, definitions and primary sources over marketing copy. Every page needs statements that can be quoted without context.
Structured data
Organization, Service, FAQ and Article schema turn text into machine-readable entities — the basis on which AI systems derive relationships and trust.
Machine readability
AI crawlers must be allowed in and able to read content without JavaScript rendering — via robots.txt, llms.txt and Markdown delivery instead of rendered storefront fragments.
Entity & earned media
Consistent company data, directories, reviews and mentions on third-party sites weigh more than your own brand copy — the AI trusts what others write about you.
Direct answers
Every question a buyer could ask gets a self-contained answer in 40 to 60 words — as a heading, FAQ or intro. That is the unit AI systems pick up.
From crawl check to citation.
Bot access (robots.txt, firewalls, bot management), llms.txt, structured data, renderability and a coverage check against the questions your buyers actually ask.
Schema.org graph with Organization, Service, FAQ and Article, llms.txt and llms-full.txt, Markdown delivery for AI crawlers, clean canonicals and hreflang — implemented in Shopware code.
Category, service and guide pages with direct answers, primary sources and FAQ blocks — structured by the questions from the coverage check, not by keywords.
Tracking AI crawlers in the server log, recurring citation and coverage checks, llms.txt and sitemap maintenance with every new piece of content.
Four findings from the research.
Others have to write about you
Studies show that AI systems weight mentions on third-party sites — directories, trade media, reviews — more heavily than your own website. GEO therefore does not end at the edge of your domain.
Every engine ranks differently
ChatGPT, Perplexity and AI Overviews use different indexes and crawlers and prefer different source types. There is no optimization "for AI" — only per system.
Neutral beats promotional
LLMs prefer factual, evidence-based content. Superlatives and marketing phrases lower the probability of being cited — numbers, definitions and sources raise it.
Blocked bots don't cite
Many shops block GPTBot, ClaudeBot or PerplexityBot through bot management or CDN rules without knowing it. What isn't crawled can't be cited — the check takes minutes.
We do this ourselves — on this website.
The best reference for GEO is a website that AI systems actually cite. Our own is our test bed.
What we recommend, we run here: every page is delivered to AI crawlers as
Markdown on request, llms.txt and llms-full.txt are refreshed with every
build, the Schema.org graph links organization, locations, people, services
and FAQs, and we track in the server log which AI assistants retrieve our
content. Alex Jank summarized the research behind it in his talk
"Beyond SEO: Why AI Search Needs a New Playbook"
at SCUC 2026. How AI systems in turn become a sales channel is covered in our
article on AI sales channels with Shopware.
And because GEO stands on the same technical foundation as classic SEO, we
implement both in one step — see Shopware SEO.
GEO pays off when customers research before they buy.
- Anyone looking for a guaranteed
- Shops unwilling to rework content on a factual basis.
- Pure price leaders without content substance.
- Merchants and manufacturers whose customers research complex products.
- B2B suppliers whose buyers increasingly pre-select vendors via AI assistants.
- Shops with real content substance that don't show up in AI answers yet.
Frequently asked questions about GEO and AI search.
What is Generative Engine Optimization (GEO)? +
GEO refers to measures that get a website named as a source in AI-generated answers — for example in ChatGPT, Perplexity or Google AI Overviews. Instead of rankings, GEO targets citation and covers bot access, structured data, fact-based content and consistent entity signals.
How is GEO different from SEO? +
SEO optimizes for a results list, GEO for a generated answer. The technical foundation is the same — crawlability, performance, structured data. GEO adds machine-readable formats such as llms.txt, citable direct answers and third-party mentions, which AI systems weight more heavily than your own brand copy.
How does my shop get cited in ChatGPT or AI Overviews? +
By letting AI crawlers in, making content readable without JavaScript, delivering concrete facts and answers on each page and making your company recognizable as an entity — via Schema.org, consistent NAP data and reviews. Which sources are preferred differs per engine, which is why we check citations per system.
What is an llms.txt? +
An llms.txt is a text file in the root of a website that gives AI systems a curated overview of its most important content — similar to a sitemap, but in Markdown and with context. The llms-full.txt variant contains the full content for crawlers that don't fetch pages individually.
Should I block AI bots like GPTBot? +
Only deliberately. Blocking GPTBot, ClaudeBot or PerplexityBot protects content from training — but forfeits being cited in answers. For merchants, the visibility benefit usually outweighs the cost. What matters is that the decision is made consciously and not by a CDN or bot-management default profile.
How do I measure visibility in AI search? +
On three levels — AI crawler hits in the server log, recurring citation checks with real buyer questions per engine, and referral traffic from ChatGPT, Perplexity and others in analytics. We set up the tracking and check monthly whether new content is retrieved and cited.
Does GEO work specifically for Shopware shops? +
Yes — Shopware 6 provides clean URLs, sitemaps and meta data, but no llms.txt, no Markdown delivery and only basic structured data. We close these gaps in code and structure category and product content so it works as answers.
What does GEO cost? +
The GEO audit is part of our SEO or system audit. Technical implementation is quoted as a project based on the measure list, ongoing content and monitoring work as a retainer in the four-figure monthly range — depending on scope and competition.
Is your shop being cited?
We check bot access, structured data and coverage against real buyer questions — and show you what's missing. Fact-based, with an honest effort estimate.
