Anyone searching today for a product, a service or an answer increasingly gets a ready-made text from Google before the first classic search result even appears. These AI summaries, known as AI Overviews, condense several web pages into one answer and list the sources used below it. In March 2026 they appeared on around 48 percent (BrightEdge) of all searches, a clear jump from the roughly 20 percent (Ahrefs) before. For every business website this shifts the decisive question: is it still enough to rank on page one, or does my website itself have to appear as a source in the AI answer? This article explains what changes for visibility in 2026, how websites end up in AI answers at all (the topic of GEO) and which concrete steps companies should take now. No one can seriously promise a fixed ranking, but a well-considered approach noticeably improves the starting position.
What AI Overviews Change About Visibility
AI Overviews are the AI-generated answer boxes Google shows above the classic ten blue links. Instead of a list of results, Google delivers a cohesive text assembled from several websites and links the sources below it. Whoever appears there as a source gains visibility and trust before the user even scrolls. Whoever is missing loses attention to the cited competitors. An AI Overview does not link a single page but on average 3 (Semrush) different sources per answer. So it is no longer about first place alone but about belonging to the circle of cited sources at all.
A second effect is decisive economically: the zero-click search. When the AI answer already answers the question, a share of users no longer clicks on any result. Analyses show that the click rate of the first organic position drops by around 58 percent (Ahrefs) once an AI Overview is displayed. At the same time Gartner forecasts that classic search engine usage volume will fall by about 25 percent (Gartner) by 2026 as queries shift to AI assistants and answer engines. This shifts the value of visibility: it is no longer enough to rank somewhere on page one; you have to appear in the answer itself or be linked there as a trustworthy source. How to set up your website for findability in general is described in our article on the SEO basics for businesses.
Briefly explained: GEO and AEO
Why Zero-Click Shifts the Rules
The classic logic was: rank well, earn the click, bring visitors to the website. In AI search an intermediate step appears. Google reads several websites, formulates its own answer and shows it at the top. The user often gets their information in the answer box without opening a single page. For websites this means two things. First, direct traffic drops for purely informational queries because the question is already answered in the overview. Second, the value of a visible mention rises: whoever is named and linked in the answer benefits from the attention and from an implicit vote of confidence, since the AI has evidently judged the page reliable.
For companies this does not make classic SEO obsolete, quite the opposite. The sources an AI Overview draws on come predominantly from the top organic results. Whoever ranks well in the classic sense therefore has a far higher chance of being cited in the AI answer too. The two worlds are connected. The type of query also matters: transactional and commercial searches, where someone wants to buy, book or enquire, so far less often trigger a pure text answer, because concrete providers, prices and reviews are what count here. These very searches are the revenue-relevant ones for most websites, and here the click remains the norm for now.
Why Now Is the Right Time
Citable Content Instead of Scattering Keywords
AI systems rarely read whole articles in one go; they pull out individual paragraphs, lists or tables. Content must therefore be modular: every paragraph should be understandable in isolation and answer a concrete question. The most effective format for this is genuine FAQ content, where a naturally phrased question stands as a heading with a concise, complete answer directly below it. The question should sound the way a person actually asks it, for example How much does a new website cost or How long does a website relaunch take. Strings of keywords without an answer, by contrast, hardly help an answer engine.
This format pays off twice because many searches now run through voice and contain naturally phrased questions. Estimates suggest that around 58 percent (CXL) of voice searches target local or business information such as opening hours, address or availability. A website that answers exactly these questions in clear language thus serves two channels at once: classic search and the AI or voice answer. The decisive difference is in the detail: not the company's jargon counts but the language of the customers. How to build content deliberately is explored in our article on conversion optimization for more enquiries.
Question as Heading
Every FAQ starts with a naturally phrased question, the way customers type or speak it, including service and context.
Direct Answer
The first line answers the question completely. Background follows afterwards. This lets the AI cite the core cleanly.
Modular Blocks
Paragraphs, lists and tables that are understandable on their own. Each block carries by itself, not only within the whole text.
Technical Foundation and Schema Markup
Schema markup is a standardized annotation in the source code that makes content unambiguous for machines. It tells a search engine not just that text is here, but that this is a company, this is its address, this is a question with an answer, this is an article with an author and a date. Without this structure, content is less clearly captured by AI systems. The effect can be judged realistically, though: a controlled study of 1,885 (Ahrefs) pages that added JSON-LD markup between August 2025 and March 2026 found no notable increase in AI citations compared to reference pages. Schema markup is therefore not a switch for more citations but a machine-readable foundation that makes a page easier to recognize and attribute. On top of that, only about 23 percent (Schema.org) of German SMEs use structured data at all. That gap is a concrete opportunity for carefully built websites.
For most business websites several markup types are especially relevant: Organization or LocalBusiness describe the company with name, address and contact, FAQPage marks up questions and answers machine-readably, Article labels blog posts with author and date, Product and Offer describe offerings. All of this can be embedded cleanly as JSON-LD in the page head without changing the visible design. It is important that the markup mirrors exactly what actually appears on the page, since misleading annotation violates the guidelines. The technical foundation beneath it counts just as much: a fast, clean website is the prerequisite for being considered a source at all. How to improve load time and stability is shown in our article on the Core Web Vitals, and the technical implementation is part of our SEO work.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "How much does a new website cost?",
"acceptedAnswer": {
"@type": "Answer",
"text": "The price depends on scope, number of pages and features. A solid business website starts in the four-figure range; complex projects go beyond that."
}
},
{
"@type": "Question",
"name": "How long does a website relaunch take?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Typically a few weeks to a few months, depending on content scope and approval process."
}
}
]
}Schema Markup Checklist
Trust Signals and a Clear Entity
An entity is the digital identity of a company: who we are, what we offer, how to reach us, what we stand for. AI systems and search engines assemble this identity from many signals, among them consistent details about name, offering and contact across all channels, mentions on third-party sites, reviews and a recognizable subject-matter authority. Contradictory or outdated details weaken this signal and make it harder for the AI to attribute different findings to the same company. For locally active businesses, consistent NAP data also matters, meaning name, address and phone number spelled identically everywhere online. How the Google Business Profile serves as a foundation is shown in our article on local SEO and the Google Business Profile.
Data quality and reputation are not a side issue but directly revenue-relevant. Around 62 percent (BrightLocal) of consumers avoid companies with incorrect or contradictory details, and about 71 percent (BrightLocal) turn away from providers with a rating below three stars. It is also notable that around 32 percent (BrightLocal) of consumers already consider AI tools superior for research tasks. A clear, well-maintained entity with consistent data and good reviews is therefore at once a ranking signal, a trust anchor for people and an identifying feature for machines.
| Aspect | Weak setup | Strong setup |
|---|---|---|
| Content | Long running text without clear questions | Modular, FAQ, question-answer structure |
| Machine readability | No or faulty schema | Clean JSON-LD matching the content |
| Entity | Contradictory details online | Name, offer, contact consistent everywhere |
| Trust | Few, unanswered reviews | Current, maintained, answered factually |
| Tech | Slow, unstable, outdated | Fast, stable, regularly maintained |
What Companies Should Do Now
The right order is decisive: GEO does not replace classic SEO but builds on it. First the foundation must be in order, meaning a technically clean, fast and well-structured website with a clear information architecture. On this basis citable content, FAQ and schema markup unfold their effect. Then it is worth systematically collecting the most important customer questions and writing a modular, directly answering section for each. In addition, reviews should be actively gathered and answered factually, and details should be kept consistent across all channels. Anyone wanting to know what budget such measures realistically require will find orientation in our article on what a website costs in 2026.
A word on expectations belongs here. No one can promise a specific position in the AI answer or in the classic search results, because Google calculates both from many signals and adjusts the systems continuously. What can be influenced are the factors that are demonstrably assessed: relevance, clear structure, trust signals and a clean technical foundation. From more than 50 website projects (project experience) we know that continuous, clean work carries further in the long run than short-term tricks. The good news: almost everything that makes websites better for AI search also makes them better for people, clearer, faster and easier to understand.
- Secure the technical foundation: load time, structure, mobile display
- Collect customer questions and write a direct, modular answer for each
- Embed FAQPage, Organization and Article schema cleanly as JSON-LD
- Keep details about name, offering and contact consistent across all channels
- Actively gather reviews and answer them factually
- Classic SEO first, then extend the AI layer deliberately