GEO vs. SEO - similarities & differences

What is Google AI Mode?

Google AI Mode ist ein neuer experimenteller Modus in der Google-Suche, der im Frühjahr 2025 (vorläufig nur in den USA) lanciert wurde (Google, 2025).

Instead of just a list of blue links, AI Mode provides AI-generated answers and goes one step further than previous AI searches such as ChatGPT.

This is because rich results such as images, Google Maps maps, videos etc. can also appear within the AI answer pages, which makes the whole user experience much more interactive and versatile than in previous AI chatbots, which mainly answer in pure text format.

In contrast to traditional Google search, AI Mode is dialogue-oriented and relies heavily on personalization. The past search history should be contextually incorporated into the answer for each search query, which means that results can differ greatly from user to user depending on their preferences and interests for the same input.

AI Mode can also process a wide variety of input types, making it multimodal. You can enter questions by voice memo, type them in as text or upload images.

Google's latest AI(Gemini 2.0) is responsible for the answers, allowing even complex questions to be contextualized and answered(Google, 2025).

Here you can take a first look at what Google AI Mode looks like in action.

https://www.youtube.com/watch?v=qbqZQFOVfA8

Why LLMs are the future of search?

Well, if we're honest, the whole thing doesn't really come as a surprise.

Most of us have probably been aware for some time that the future of search will look different than a list of blue links.

The introduction of Large Language Models (LLMs) offers a much faster, personalized way of compiling information and mapping it to the specific search query.

In the beginning, LLMs still had the limitation that they could not access live data. However, ChatGPT, Perplexity & Co. can now also access live data and thus provide even better answers.

Google was forced to change the technology of its important product due to competitive pressure from new players such as ChatGPT.

This is because generative searches offer the user a better user experience in many respects. And if Google wants to maintain its high market share, it has to move with the times.

56% of Google's 350 billion revenue in 2024 came from advertisements in Google Search. If Google loses market share, this would result in major financial losses.

SEO 2.0 - What remains and what changes?

The question now arises for marketers, SEO agencies and companies alike:

How do we deal with this change in the search?

Which strategies are still proving their worth in the age of AI and which are yesterday's news?

Wie kann man Websites für KI-Suchen optimieren?

What remains unchanged?

Not all SEO measures no longer play a role in LLMs - as various studies have shown.

In his study, SEO expert Tomasz Rudzki used 25,000 different search queries to investigate how closely Google search results matched links in AI searches.

The result: If you rank number 1 on Google, you have a 25% chance of being cited by AI searches.

This means that many proven SEO basics are still a basic requirement in the age of AI.

Your website must be technically clean (clean code structure, fast loading times, mobile optimization). High-quality content with real added value remains the be-all and end-all - perhaps even more so than before, as AI prefers trustworthy, relevant content.

Backlinks and the authority of your domain are still important, even for LLMs.

What is changing with AI search?

Many refer to SEO in the age of AI searches as GEO - Generative Experience Optimization. The aim is to design content in such a way that it is attractive to AI.

GEO has two goals: to be cited as often as possible and to make your brand, product or service directly visible in the AI response.

Ato Herzig, Co-Founder Beyondweb GmbH

For example, if someone is looking for "the best newsletter software", then it is advantageous if your brand/product appears in the answer (in the best case even linked).

This does not necessarily lead to a click, but it increases your chances that the user will consider your offer for a purchase.

Here is an overview of the biggest changes from GEO to traditional SEO:

Citation-Friendly Content

A new trend that is emerging is "optimizing for citations".

In LLMs and AI Overviews, being quoted counts for a lot.

Links do not necessarily lead to a click, but it increases your chances that the user will consider your product/service for a purchase.

End of the "Corporate Wikipedias"

Die Tatsache, dass der AI Mode und ChatGPT generische informelle Suchanfragen längst schneller und personalisierter beantworten können, als jeder Blogartikel heisst: Das Zeitalter von generischen Lexikonartikeln ist vorbei.

But beware: I'm not saying that there's no point in writing high-quality blog articles anymore.

There is still a demand for "slow food" - people who actually want to read an article on a topic.

Thought leadership, storytelling and testimonials from people can still be in demand. It can also continue to be worthwhile to publish high-quality blog articles to influence AI responses.

But the influence of informal content on your organic traffic will be much lower in the future.

Passage optimization instead of just page optimization

Google AI Mode not only indexes entire pages, but increasingly sections (passages) of them, as SEO expert John King has discovered.

This means that a single paragraph of your article could end up in an AI response without your entire page being crawled.

Make sure that your content is organized into logical, thematically focused sections.

Each section (with its own subheading) should answer a specific question or clearly address an aspect. This increases the chance that this section will be used by the AI as part of the answer.

Semantics and context

Keywords are still important, but AI pays more attention to meaning and context than to exact word matching.

Semantic SEO is becoming a must.

This means that you must fully cover synonyms, generic/subordinate terms and related topics in your texts. The AI makes semantic connections - e.g. it understands that "pet" has to do with "dog" or "cat". Content that covers topics holistically is preferred.

Your content should also precisely match the intention behind the query (informational, navigational, transactional, etc.), because like traditional search engines, AI tries to fulfill the user's intention, not just match words.

Structured data & schema markup

Schema.org & Co. were already relevant, but will become even more important in the future. They help AI to better interpret content (e.g. recipes, FAQs, product data).

Wenn Google bestimmte Fakten aus deiner Seite ziehen kann (Preise, Öffnungszeiten, Ratings) ohne die gesamte Seite crawlen zu müssen, steigen die Chancen, in einer AI-Antwort direkt erwähnt zu werden.

Multi-format optimization

Content in different formats is becoming more important.

Images should have meaningful file names and alt texts, videos need good descriptions or transcripts - because the AI reads these too.

Google AI Mode can theoretically generate part of the answer from a video transcript or a podcast quote.

AI Search requires new measurement methods

Long-proven SEO KPIs such as organic clicks or keyword rankings become less important if many search queries end without a click or rankings are always context-dependent.

We are still at the very beginning - suitable tools and measurement methods are still being developed.

Another important point: Google currently provides hardly any official data on when or how often you appear in AI results - this makes monitoring more difficult.

Nevertheless, there are already initial indications of the direction that performance measurement will take.

Prompt share

This term is understood as "share of AI queries". It is about measuring the percentage of relevant AI search prompts in which your brand is mentioned as part of the answer.

Whereas previously the impressions in Google Search Console showed how often your page was in the top 10, now it's about how often you appear in AI responses.

So far, you can only estimate your prompt share, e.g. by testing important keywords: Does your page appear in the generated responses? In the future, tools will measure this value. A high prompt share means that your content is indispensable for the AI and is often cited.

LLM visibility

LLM stands for Large Language Model (i.e. the AI behind the search). LLM visibility measures how visible your brand is in a specific language model.

Simply put: How well does the model "know" your content?

This is shown by whether it quotes you, mentions you or links to you. You can increase this visibility by building authority (good content, mentions or links from external domains, etc.), because the AI model has been trained with masses of web data - well-known brands/sites have a trust advantage.

In other words, LLM Visibility is the new form of Domain Authority.

Quality Traffic & Engagement

Google itself already advises paying more attention to the quality of visits instead of the number of clicks. If users reach your website via AI Mode or ChatGPT, they are often further along in the funnel (because they have already researched a lot of information in advance without clicking on the website).

New KPIs are therefore things like the conversion rate of AI searches (do these less frequent visitors perform a desired action more often?) or dwell time (do they stay to read more details because the AI search has hooked them?).

Diese Metriken sind aktuell nur schwierig auswertbar, aber diese Engagement-Metriken könnten trotz weniger Traffic stabil bleiben oder sogar steigen.

Use new tools

The SEO community is already developing tools to better understand AI Search. AlsoAsked (known for People Also Ask questions) continues to help find thematically related questions - which is exactly what you can use to provide content for fan-out queries.

SEOs are also experimenting with tools like ZipTie or WordLift's Query Simulator to see what questions an AI might trigger. Look out for features in popular SEO tools: Many will soon offer AI Visibility Reports that show which prompts you show up for.

In short: Use everything that gives you a measurement advantage and be prepared to try out new tools apart from Ahrefs, SEMrush & CO.

By running your own tests (regularly entering typical search queries yourself on ChatGPT), you will get a good feeling of how well you are represented in AI searches.

There are also more and more AI visibility tracking tools such as Peec.ai or Profound. However, these tools are still in the beta phase and are not yet fully developed.

If you have access to Google Search Console, you can also build your own AI traffic dashboard in Looker Studio.

In general, however, it is important to understand that the organic click curve tends to decline, but there can still be success with SEO, for example through mentions or links in the AI responses.

This means that providers need to rethink their approach: visibility without a click can be valuable. The "share of voice" is shifting to the world of AI results.

This means that providers need to rethink their approach: visibility without a click can be valuable. The "share of voice" is shifting to the world of AI results.

Ato Herzig, Beyondweb GmbH

This means that providers need to rethink their approach. Visibility without a click can be valuable.

The days when organic traffic was the be-all and end-all are definitely over!

5-step plan for companies

So how should you go about preparing for AI Search? Here is a 5-step plan that you can implement right away:

Perform status quo audit

Get an overview first. Analyze your current rankings and content: Which top keywords are you currently ranking well for - and are AI overviews already appearing for these searches?

Check in Search Labs (if available) or with screenshots/reports from the industry whether your website is cited in AI responses. Also examine technical basics: Is your site crawlable, fast, mobile optimized?

This SEO foundation must be right, otherwise you have no chance in either classic or AI results.

Adapt content strategy

Identify the most important questions that your target group asks and expand or design your content accordingly. Conduct a topic & question research (e.g. with AlsoAsked or Google Questions).

Create content that answers these questions directly - preferably in dedicated sections or via FAQs. Use clear language (B1 level 😉) so that the AI can easily interpret your statements.

Also plan different formats: Texts with helpful images, infographics, videos or audio, if appropriate. Keyword holistic: Be the best one-stop content on a topic so that the AI can't help but use you as a source.

Implement on-page & technical SEO 2.0

Revise critical pages according to the new optimization principles. Use structured data wherever possible (FAQ page, HowTo, Product, Article, etc. depending on the content).

Build important facts into good HTML elements one - e.g. <ul>-Lists for enumerations, <table> for overviews - so that the AI can easily extract the information.

Check your meta data: Title and description should concisely summarize the content (they could serve as an AI tooltip). Keep technical performance high and eliminate errors that hinder rendering. In short: make it easy for the AI to find and utilize your content.

Track new KPIs & train the team

Set up a dashboard that not only shows classic SEO KPIs, but also new indicators.

Examples: Number of featured snippet/AI source mentions(collected manually), changes in organic traffic in the context of AI rollouts, conversion rates of traffic. Educate your team on the importance of prompt share and LLM visibility - explain why traffic may be decreasing but brand visibility is increasing.

Everyone involved (content, PR, management) must understand that AI visibility is a success, even if it is not immediately measured in clicks. This step is important in order to create internal acceptance for the new measures.

Introducing a testing loop

Go into continuous test mode. AI search is constantly evolving - what is true today may be different in 6 months. Establish a process to experiment regularly: e.g. run a prompt test once a month for your top topics (ask the AI about your products/services - do you show up?). Test different content formats and observe whether and how they are picked up by the AI. Carry out A/B tests with content blocks (e.g. install a more concise answer box) and see whether this has an effect on rankings or AI citations. Get feedback from your users: Do they find the AI answers helpful, are they missing information? All of this feeds back into content optimization. Learning by doing is the motto here - agile testing is the quickest way to learn what works.

Regardless of whether you follow the 5-step plan above exactly or not.

It is important to start now!

Even if AI Mode only accounts for a small proportion of traffic today, you will be in pole position if you adapt your SEO strategy early on.

If Google AI Mode suddenly launches in Europe sooner than expected and becomes the new default search (which I think is likely), then you can't start early enough.

My conclusion & outlook

Google AI Mode brings challenges, but also opportunities. Especially for those who adapt to it early on.

Companies that are willing to learn and adapt will also be visible and successful in the age of AI.

It is important to set the course now, to take advantage of the new opportunities and to understand SEO as a dynamic field that constantly brings innovations.

If you want to know whether you are exploiting your current SEO potential, request our free SEO potential analysis - find out how well equipped your site is for AI Search and where untapped potential lies dormant.