The AI answer belongs to those who shape it

Generative Engine Optimization is not an SEO update – but an economic stimulus package for PR. Why communicators now have the strongest cards.

From Martin Regnet

Take the test: Open ChatGPT, Gemini or Perplexity and ask for the most trustworthy providers in your industry. Does your company appear? In what words? In what context – as a solution provider or as a problem case? And if it doesn’t appear at all: That’s not a bug. It’s the new reality.

I carried out this reality check live with the participants of our AG CommTech webinar in March 2026. The result was a wake-up call for many. Because if you rank on page 1 of Google, you can still be invisible in AI answers. The rules of the game have changed fundamentally – and the answer is not “more SEO”, but Generative Engine Optimization (GEO).

The end of the traffic economy

SEO was a traffic economy: whoever generated the most clicks won. GEO is a reputation economy: whoever is the most credible, most quoted voice in the subject area wins. That sounds like nuance. But it is a paradigm shift.

The figures are clear. 65 percent of the German population regularly use AI language models. Google’s market share in Germany has fallen below 90 percent for the first time. The zero-click rate at the beginning of 2026 is 75 percent – three out of four search queries end without a single click on a website. For queries with Google AI Overviews, the organic click rate drops by up to 58 percent.

What does that mean in concrete terms? Every page that used to get 100 clicks now only gets 42 on average – with the same SEO ranking. And this loss of reach is barely visible in traditional Google Analytics. It happens silently.

Why PR has the strongest cards now

Now comes the good news. Large language models do not rely on advertisements when generating answers. They rely on journalistic media, studies, expert quotes and thought leadership formats – classic PR tools. These are not SEO artifacts. These are not performance marketing campaigns. This is the terrain of corporate communications.

The consequence is radical: anyone who is not mentioned in credible external sources practically does not exist for the AI response. The model is not malicious – it simply cannot find the signals it wants to rely on. PR thus becomes the infrastructure of AI visibility. Not an accessory, but a prerequisite.

Those who do not set their own narratives are controlled. AI fills every vacuum – with old LLM training data, critical reports or competitor narratives. In customer projects, I have seen ChatGPT reproduce incorrect founding dates, name CEOs who have long since been replaced or let frustrated Glassdoor reviews shape the image of a company. Not an exceptional case. Everyday life.

How AI really thinks – and what that means for content

LLMs work fundamentally differently to search engines. Search engines index and rank according to keywords. LLMs calculate probabilities. They do not ask how often a word occurs, but rather: What does this company say about a topic – and what do others say about it? GEO is therefore not a switch that you flip. It is probability management.

Fraunhofer IESE has broken down the LLM process into four steps: Retrieval, Ranking, Synthesis, Generation. There are specific levers in each step. When retrieving information, presence in quality-assured sources counts. In ranking, credibility and E-E-A-T conformity are decisive. During synthesis, lines of argumentation from PR texts flow directly into the answer. And consistent brand mentions act as anchor points during generation.

The decisive factor here is that there are two ways into the AI response. Retrieval augmented generation (RAG) accesses live sources – what counts here is topicality. Parametric knowledge comes from the training data set and is typically six to eighteen months old – long-term presence in high-quality media is what counts here. Both must be used.

Three layers determine your AI image

AI reputation – the picture that generative systems paint of your company – is made up of three layers. The factual level: year of foundation, locations, key financial figures, facts about the company’s focus. Surprisingly often incomplete or outdated. The narrative layer: What topics does the AI associate your company with? Innovation, growth – or job cuts and greenwashing? And the context level: In which competitive landscape does AI place you?

The good news is that all three layers can be influenced. The E-E-A-T framework – Experience, Expertise, Authoritativeness, Trustworthiness – provides the compass. A CFO who publishes a personal experience report from the Capital Markets Day sends a strong experience signal. A position paper with an industry association strengthens expertise. Media quotes and Wikipedia entries build authority. And consistency across all channels – from the CEO bio to the number of employees – ensures trustworthiness. Any inconsistency is a negative signal that makes LLMs turn to third-party sources.

Get started in four steps

GEO does not need a special budget or a new team. It needs a system. Four steps make it possible to get started:

1 – Reality check. Ask ChatGPT, Perplexity and Gemini about your company and your industry. Document systematically: Where do you appear? With which words? How do you compare to the competition? Where are there narrative gaps or misinformation? Tools such as peec.ai or Cuemarc enable in-depth analyses – but a structured prompt library also provides valuable insights.

2 – Set fact anchors. Without clean, consistent facts, AI builds on gaps. Basic data, product descriptions, expert profiles and three to five core messages must be formulated identically across all channels. No channel should broadcast a different core identity. A clear messaging house and factsheets are not an optional extra, but a must.

3 – Build content infrastructure. AI models prefer certain formats: FAQ pages with a clear question-and-answer structure, definitions in the format “X is…”, how-to-guides, schema markup. The basic rule: Each core statement must exist as an independent, clearly formulated piece of text that an AI can extract directly.

4 – Maintain the source ecosystem. AI gives more weight to information that is confirmed by several independent sources. Earned media, own studies, Wikipedia entries and industry publications form a network that uniformly confirms the same expertise. Targeting those niches in which competitors are not yet present. GEO does not end with your own website – it is classic positioning work with new strategic leverage.

Start now

AI visibility is not an SEO appendage. It is a field of reputation in its own right that directly affects corporate communications – and that plays right into our hands as communicators. The core narratives are being set now: by you or by others. The best time was yesterday. The second best is today.

GEO workshop at the CommTech Academy

On April 10, 2026, I will be offering a four-hour online workshop that goes into all the blocks of the GEO framework in greater depth – including measurement methods, KPIs and an individual 90-day roadmap. Further information at commtech-academy.de.