Contents
    Comparison: traditional search results list and a cited AI answer, connected through GEO
    From the results list to the cited AI answer.

    Generative Engine Optimization (GEO) is the practice of optimizing so that AI systems such as Google AI Overviews, ChatGPT, Gemini and Perplexity name a brand and cite its content. 68% of Google searches in the US end without a click according to SparkToro/Similarweb (as of 2026), ChatGPT counts around 900 million weekly users, and in Switzerland four in five people aged 15 to 75 already use AI according to the IGEM Digimonitor 2026. Visibility increasingly happens inside the answer itself, not in the results list.

    For businesses, this shifts the goal: ranking number one is no longer enough. You need to be the source the AI builds its answer from, or be on the page it reads to build it.

    What is GEO?

    GEO covers every measure that gets a brand into AI-generated answers: as a recommendation ("Which fiduciaries in Zurich can you recommend?"), as a mention, or as a cited source. How a recommendation comes about and which levers make it more likely is covered in How do I get recommended by ChatGPT?. The term comes from a Princeton study from 2023, which showed that statistics, source references and citable structure raise visibility in AI answers by up to 40%. Related terms are AEO (Answer Engine Optimization, answering one specific question) and LLMO. In practice they all mean the same thing: being found when someone asks an assistant instead of Google.

    How does GEO differ from SEO?

    SEO optimizes for people finding your page through the results list. GEO optimizes for machines understanding, trusting and citing your content. Both share the foundation (crawlable, structured, credible content), but GEO additionally works on the sources outside your website and on the models' memory.

    AspectSEO (traditional)GEO (AI search)
    GoalClick on the pageMention or citation in the AI answer
    UnitPage / keywordStatement / entity / prompt
    SuccessRanking, clicksMention rate across a prompt set, citations
    FormatTitle, meta, contentClear answers, prices, FAQ, structure
    Sourceyour own pageown page, third-party pages, the model's memory
    Competitionten blue linksthree to eight names per answer

    The most important difference is the last row: an AI answer compresses the market to a handful of names. If you are not among them, you do not exist for that query.

    Why isn't traditional SEO enough in 2026?

    Because the assistants keep their own shortlist. In our AI visibility study of 1,515 AI answers and 14,135 traced mentions across 12 Zurich industries, roughly three quarters of the recommended firms were not in Google's top 10, even for the exact searches the assistants ran themselves. A good ranking is the ticket in, not the selection.

    Add the click loss: when 68% of US searches end without a click, you lose reach the moment your brand doesn't appear in the answer. GEO makes sure you get named there.

    How do you get cited in AI Overviews and ChatGPT?

    We measured separately which searches trigger an AI Overview in Switzerland and which websites Google cites there: Google AI Overviews in Switzerland.

    AI systems favor pages they can easily excerpt and substantiate. Our study compared the own pages that were cited with those retrieved but ignored: pages with an FAQ section are cited 1.6 times as often, pages that name Zurich explicitly 1.35 times. Hence:

    • Answer first: Answer the central question in the opening sentences (who, what, for whom, where, from what price).
    • Organize by questions: Phrase headings as real user questions, the way customers put them to an assistant.
    • Concrete details: Prices, formats, duration, location, numbers. Assistants often lift such sentences verbatim into their recommendation; claims like "tailored" and "holistic" they do not.
    • Self-contained paragraphs: Every paragraph should make sense without context, because it gets cited on its own.
    • FAQ section: Short, precise answers to the questions that come up in a first call.
    • Structured data: Organization, Service or LocalBusiness, FAQPage and Article schema, with price data where possible.
    • Access for AI crawlers: Don't block GPTBot, ClaudeBot, PerplexityBot and Google-Extended. A blocked page cannot be cited.
    • E-E-A-T: Make author, date and expertise visible.

    Is optimizing your own website enough?

    No. That is the most common false assumption in GEO guides, and our data contradicts it on two counts.

    First, a large share of mentions has no source at all. About 40% of the 14,135 mentions in our study came from the model's memory: the assistant already knew the firm without opening a page while answering. Across all twelve industries, 39% of mentions relied on the firm's own website and 18% on third-party pages. The split varies a lot by industry: for dentists, 50% of mentions come straight from the practice website, for physiotherapy 69%, for wealth managers 57% come from memory. What to do first therefore depends on the industry. Google AI Mode is a special case: there, three in four answers rely on Google Maps, as our measurement One year of Google AI Mode in Switzerland shows.

    Second, which page you are on matters more than how good you are. Among firms on a list page the assistant demonstrably read, the stronger half was picked 42% of the time, the weaker half 32%, a factor of 1.3. Being on that page at all is worth 8.2 times as much. And within the page, firms in the top third are named 31% of the time, firms in the bottom third 21%. Moving up a list usually costs an email, not a budget. For broader markets, AirOps puts brand mentions from third-party sources at 85%, almost all of them lists, comparisons and reviews.

    On top come two thresholds the study made visible: the share of named firms rises from 67% at 1 to 9 Google reviews to 91% at 30 to 99 reviews, then flattens; the star rating itself barely separates named from unnamed firms. For referring domains the threshold sits at around 99, beyond which more of the same changes almost nothing. What this looks like for your industry is in the study, with lever tables for all twelve industries. Where you stand today is shown by the free AI visibility check.

    How fast does GEO work? An example on ourselves

    The domain hierarchy.ch went live in August 2026. On 8 September 2026 we asked ChatGPT, Gemini and Claude, each with web search, 14 questions customers ask about AI workshops in Zurich. ChatGPT named hierarchy.ch in 11 of 14 questions, Gemini in 6, Claude in 2, next to providers that have ranked for those terms for years. All three quoted the wording of our workshop page verbatim: formats, deliverables and the review after 30 days. In Google, the same pages still sat at positions 50 to 90 for the matching search terms.

    That shows two things. Assistants with web search judge a page by its content when they retrieve it, not by domain age; concrete details work as soon as the page is indexed. The route through the models' memory, the 40% of mentions with no source opened, does not get shorter because of it: it builds over months from mentions in third-party sources and only becomes visible through repeated measurement.

    Checklist for GEO-ready content

    1. Answer the core question in the first 2-3 sentences, including location and starting price.
    2. Phrase headings as questions people actually ask.
    3. Concrete details instead of claims: formats, duration, deliverables, numbers.
    4. Add a table or list for comparable information.
    5. Include an FAQ section with short, precise answers.
    6. Serve structured data: Organization/Service, FAQPage, Article, with prices.
    7. Allow AI crawlers.
    8. Show authorship, date and sources.
    9. Get onto the list pages and directories of your industry that the assistants demonstrably read, and as high up on them as possible.
    10. Build your Google review profile towards 30 reviews; beyond that the effect flattens.
    11. Set up a fixed prompt set and measure monthly whether your mention rate is rising.

    How do I measure GEO success?

    Not by rankings. The unit of measurement is a fixed set of real customer questions, put repeatedly to ChatGPT, Gemini, Claude, Google AI Overviews, AI Mode and Perplexity. For each answer you record: named, cited or merely retrieved, at which position, which sources were cited instead. Supplement that with the AI assistant channel in Google Analytics 4 and Google's Search Console report on generative AI features. How such a setup is built is described in Measuring AI visibility.

    GEO is part of a larger system of technology, content and distribution, just like good process automation. We run it as part of our GEO agency service: website, third-party sources and measurement from one team. The packages are on the service page.

    Want to know whether ChatGPT names you today? The AI visibility check is free, or let's talk.

    If you would rather not build this yourself: our comparison of SEO, GEO and AEO agencies in Zurich shows what the market charges and which questions to ask in a first meeting.

    Sources

    Tenzin Langdun

    About the author

    Co-Founder of Hierarchy · SEO, GEO & AI Expert

    Tenzin Langdun is a co-founder of Hierarchy, where he leads SEO and GEO: how visible companies are in Google, in Google's AI answers and in assistants such as ChatGPT, Gemini and Claude. He also leads the agency's own data studies, including the Zurich AI Visibility Study of 1,515 AI answers about 5,110 companies. He brings over 10 years of marketing experience across strategy, paid acquisition and SEO, and has worked at KPMG, EY, Siemens, Adnovum and LGT, in AI, cybersecurity, audit and consulting, and in software engineering for the insurance sector and for banking and finance. He holds a BSc from ZHAW and an MSc in Artificial Intelligence from the University of Bath. His research on AI-based thyroid cancer detection was published in Scientific Reports (Nature Portfolio) in 2024 and won the Siemens Excellence Award 2024 and the ZHAW Lab Sciences Award 2024.