Contents
Generative engine optimization (GEO) covers everything a company does to get its offer mentioned and linked in the answers of AI assistants and AI search. More and more people ask ChatGPT, Perplexity, Gemini, Claude or Google’s AI overviews instead of clicking through search results. This article explains how these systems choose sources, how GEO differs from classic SEO and which steps work in practice.
How AI assistants arrive at their answers
For GEO, it is important to understand the two sources a language model draws on:
- Training data: during training, the model processed large amounts of public text. What appeared frequently and consistently about your company there, it may reproduce even without a web search – but only as of its training cut-off and with no guarantee of accuracy.
- Live retrieval: for current or specific questions, many assistants search the web, read a handful of pages and summarise them, often with source citations. Here it matters whether your pages are found, read and judged trustworthy.
For most companies, the second route is the bigger lever, because good content and clean technology influence it directly. The first route benefits in the long run from your company being described consistently and in many places across the web.
GEO and SEO: similarities and differences
| Aspect | Classic SEO | Generative engine optimization |
|---|---|---|
| Goal | Ranking in the results list, click to the website | Mention and citation in the generated answer |
| Unit | Whole page for a keyword | Individual statements, paragraphs and facts |
| Query | Short keywords | Long, conversational questions with context |
| Trust signals | Links, user signals, content quality | Similar signals, plus consistency across many sources |
| Measurability | Rankings, clicks, impressions | Limited: referral traffic, samples of answers |
The good news: GEO does not replace SEO, it builds on it. AI search often relies on search indexes for live retrieval, so being easy to find on Google and Bing gives you a solid starting point. For apps, store listings are a separate visibility discipline, covered in our article on app store optimization.
Which content AI systems prefer to cite
Language models like to use passages that answer a question directly, precisely and in a self-contained way. That translates into concrete requirements for your content:
- Answer first: every page and every section opens with the key statement; details follow.
- Concrete facts: price ranges, project durations, processes, prerequisites, locations. Generic marketing claims such as “innovative solutions” are hardly ever cited.
- Clear structure: descriptive subheadings, lists, tables and FAQ sections make it easier to extract individual statements.
- Your own perspective: experience, decision criteria and comparisons that do not already appear on a hundred other pages.
- Freshness: a visible date, regular revisions, no outdated information.
- Credibility: clear authorship, legal notice, contact details and consistent company information.
Mentions outside your website matter just as much: industry directories, review platforms, trade articles, comparison sites and partner pages. When several independent sources describe your company in similar terms, it becomes more likely that an AI system names it as a fitting answer.
Technical prerequisites: crawlable, readable, unambiguous
- Deliver content in the HTML: many AI crawlers do not execute JavaScript, or only partially. Content that is only built in the browser stays invisible to them. Our article on SEO for web apps explains how to fix this.
- Configure robots.txt deliberately: several providers distinguish between crawlers for model training and crawlers for search or live retrieval. If you want to exclude AI training but still appear in AI search, you need to handle the respective bot user agents individually. Check the providers’ documentation, as identifiers can change.
- Structured data: schema.org markup for organisation, services, products, articles and FAQs makes facts machine-readable.
- Fast, stable pages: AI systems often fetch pages with tight time limits; slow or error-prone pages get skipped.
- Keep an eye on supplementary formats: proposals such as an
llms.txtfile that gives AI systems an overview of key content are being discussed. Their actual use by the major providers has not been reliably confirmed so far – it does no harm, but it is no substitute for good content.
Building a GEO strategy step by step
- Collect questions: what do your customers ask before buying? Sources include sales, support, Search Console queries and conversations at trade fairs.
- Assess the status quo: put these questions to various AI assistants and note which providers and sources are mentioned.
- Close the gaps: create a page or section for each important question that answers it precisely – with your facts and your perspective.
- Standardise company information: align name, services, location and contact details across your website, profiles and directories.
- Build external mentions: guest articles, interviews, customer reviews and listings on relevant comparison and industry portals.
- Repeat regularly: run the sample from step 2 monthly or quarterly with the same questions.
Measuring success – with realistic expectations
GEO is harder to measure precisely than classic SEO. AI answers vary with wording, context and user, and there are no fixed positions. Practical approaches include:
- Referral traffic: analyse visits from AI assistant domains separately in your web analytics. Some visits, however, arrive without referrer information.
- Prompt samples: query a fixed list of questions regularly and document whether and how your company is mentioned.
- Indirect signals: more searches for your brand name, enquiries saying “found you via ChatGPT” – ask for the source in your contact form.
How to choose meaningful metrics in general is covered in our article on analytics and KPIs. Stay away from manipulation attempts such as hidden instructions to AI systems in page text: they damage your credibility and can be treated as spam by search engines.
Agentic coding workshop for your team? From €799 net – request a quote by e-mailConclusion
Generative engine optimization is not a replacement for good SEO but its logical evolution: precise, fact-rich and well-structured content, technically clean pages that are readable without JavaScript, and a consistent picture of your company across the web. If your website or web app cannot be read cleanly by crawlers, we are happy to help with the implementation – with a fixed-price offer after a free analysis.
Frequently Asked Questions
What is the difference between GEO and SEO?
SEO aims for good positions in classic search results; GEO aims to be mentioned and cited in the generated answers of AI assistants. Both rely on good content and clean technology, but GEO puts extra emphasis on precise statements that can be quoted on their own.
Can I influence whether ChatGPT recommends my company?
It cannot be guaranteed, but you can increase the likelihood considerably: with content that answers specific questions, technically readable pages and consistent mentions on other trustworthy websites.
Should I block AI crawlers in robots.txt?
It is a trade-off between protecting your content and being visible in AI answers. Many providers distinguish between training and search crawlers, so you can decide specifically which access to allow.
How do I measure whether GEO is working?
Through referral traffic from AI assistants in your web analytics, regular samples with a fixed set of questions and indirect signals such as more brand searches or enquiries that mention AI tools.