From PESO to EOS: How GEO is Bringing Media Relations Back into the Strategic Spotlight in the Age of AI
A spotlight from the PR engine room in Germany on current developments in marketing
For several months now, client meetings at our firm have often begun with a new question: “How do I actually come across in ChatGPT?” What was still a curiosity a year ago is now a strategic priority. Marketing executives are feeling the pressure from the C-suite. CEOs are asking why visibility on Gemini, Claude, or Perplexity isn’t as desired in the AI responses. Why is the competitor ahead of us when people search for the best B2B tool in …? And PR agencies are suddenly at the center of a debate that has long been dominated by performance and social marketing.
Just last year, legal questions dominated the debate in Germany. What are we allowed to do? But within months, the debate has gained momentum and PR and marketing in Germany are in the midst of a conceptual upheaval – and the PR industry could play a decisive role in this.
PESO: A Brief History of the Last 25 Years
The PESO model has structured strategic communication for years: Paid, Earned, Shared, Owned. In practice, however, within the last years, it was often prioritized in a different order within the Marketing offices – more like P-S-E-O or even P-S-O-E. Paid came first, Earned fell out of focus, often in terms of budget as well. Shared fascinated clients as much as it did (and does) us.
Developments over the past two decades have favored this logic. The rise of the internet and Google (SEO and SEA) transformed advertising and the funnel. Blogs emerged, then social media, then influencers – and every new channel, every new variation, drew attention and budgets. In many places, editorial media relations was seen as something you just kept doing, but the focus, the debate, was elsewhere.
The editor’s authority as a gatekeeper was structurally weakened. Alongside the opinion leader, the influencer established themselves as the stars in Marketing. The opinion leader convinced through argumentation, whereas the influencer entices imitation through their persona.
AI is reversing this trend.
LLMs aren’t interested in packaging and style. Instead, they’re interested in the information that a critical gatekeeper posts.
EOS: The Goddess of Dawn as a Strategic Model
In Greek mythology, Eos is the goddess of dawn – she paves the way before the light appears. As a strategic acronym, EOS stands for Earned, Owned , Shared. And EOS describes the weighting logic for the world of Large Language Models (LLMs).
The reason is structural: LLMs do not aggregate advertising material. They are trained on publicly accessible, linked, cited content, that is filtered through the lens of a critical gatekeeper – that is, primarily on what traditional media relations produce. Your own website helps (really!), especially in terms of the depth and breadth of information. But critical product reviews, profiles in leading business media, and interviews carry even greater epistemic weight. AI does not distinguish based on reach, but on authority. And authority arises where independent, critical thinking happens: in editorial newsrooms, in trade publications, in well-curated communities, on Wikipedia. Thought leadership is the future. Courage is rewarded.
Paid media certainly doesn’t lose its function as an awareness tool in the funnel. The advertising column and out-of-home retain their value as catalysts. But at the decisive moment of the purchasing process, when a user asks an AI, “What are the best headphones under 200 euros?” or “Who has the best weather data for my B2B needs?”, the media budget no longer matters. And the path to AI is remarkably simple. The new trail of information gathering. Easier and faster to access than the old world of Google results. Click-click-click is a thing of the past. What matters is the narrative that independent sources have built around a brand.
This is a further fundamental shift in the balance of power: From the company and from the broadcaster to the aggregator. From the idea that markets are conversations” we have moved into an age where search itself has become the conversation.
The Third Client
At TDUB, the operational reality has changed noticeably in recent months. We’ve known the basic logic of our work for years: We always work for two clients at the same time – the client who pays us, and the editor who needs relevant input for their article. Good media relations has always been about the ability to develop narratives that stand up to an editor’s critical eye.
Now a third client has joined the mix: the LLM.
The entire process is iterative – media relations have always been that way. What’s new is the third level of iteration: We no longer optimize solely for reach and clipping quality, nor do we merely adapt to changing client dynamics, launches, etc. but rather for LLM visibility and epistemic positioning. This requires both maintaining a strategic bird’s-eye view – where does a brand stand in the AI information space? – and being active at the operational level: Which media outlets, blogs, forums, and communities need to be approached to create a consistent, authoritative brand image? Strategy comes first, followed by multiple iterations across the various levels.
Strategically, this means understanding how a company and its products are portrayed in the responses of leading language models – and why. Which sources are cited? Which competitors feature prominently? Which topics does a brand cover in the AI landscape, and which does it not? This is brand analysis taken to a whole new level.
The initial GEO findings: Earned-driven clients ahead in LLM Output
So what have we found, when initially analyzing the LLMs revealed: Our work and our key results already frequently shape the LLM output. Company profiles in business media, interviews and quotes in daily and trade media, test results – all of this already shapes the responses from ChatGPT, Gemini, Claude, Perplexity & Co. The clients who are already communicating boldly and focusing on media relations and other critical gatekeepers are already ahead; often in the lead or just behind within our GEO analysis.
What is particularly striking is the relative gap: Earned- and in some sectors owned-driven companies with well-crafted messages, backed by verifiable arguments, are usually well ahead of those that neglected media relations and owned channels and relied solely on performance, paid, and influencer marketing. The winners are those who prepare strategically, anticipate the critical thinking of gatekeepers, and incorporate it into their strategy. Because no matter where – whether in media, blogs, panels, forums, or on Wikipedia – LLMs give space to verifiable storylines.
Those who are structurally unprepared lose out. And so we were sometimes surprised at just how far ahead some of our clients already were with LLMs. Those on the client side who have already relied on clean narratives and media relations and on owned are now gaining momentum within their companies.
GEO: A closer look at today’s new workflow
Just a year ago, the clipping report was the gold standard in media relations, but the development has been breathtaking: Today, we work with specialized GEO monitoring tools for over half of our clients – platforms such as Otterly, withwhom we are in close, direct contact, learning and providing feedback on improvements to them – that enable us to systematically track a brand’s visibility and sentiment across multiple LLM systems. Based on existing materials – like SEO and SEA keyword lists, existing communication strategies, competitive analyses – and our PR experience, additional tools and analyses, we develop sets of relevant search prompts, feed them into Otterly, and analyze the results: Which sources are preferred? Are there gaps in the coverage that we can close through proactive media relations to make it into the LLMs’ relevant set? What narratives are missing from an LLM’s view of a brand? It is fascinating to be at the center of this development, which is unfolding at breathtaking speed. We are constantly learning. It’s as if we were in the middle of a new research lab and starting a degree program all over again.
GEO in Germany: Taking the plunge
The development in Germany follows a recognizable pattern. Initially, it was the legal uncertainty surrounding AI systems that drove many companies and marketing departments to adopt a wait-and-see attitude. But awareness of LLM Search has spread rapidly in recent months.
Marketing departments are electrified – right up to the top. They sense that they need to restructure themselves – and that the funnel must be conceived less technically and more narratively. New roles are emerging, and processes must be redesigned, and often in a more interconnected way. As a result, our role is also changing rapidly: We are currently in active discussions with more than half of our clients regarding GEO and LLM visibility – whether in the form of initial analyses, our new “GEO Retainer,” or the concrete, iterative implementation of PR measures derived from Otterly. We offer GEO audits and discuss the overarching LLM strategy with the client. We provide guidance on their web presence, serve as dialogue partners for CMOs on how to redefine internal structures and organizational charts, and take on new roles ourselves – often multiple ones. Sometimes strategic, sometimes providing structure at the operational level or in implementation.
This creates pressure, but also a genuine sense of optimism. The question asked most frequently in Germany is no longer “Are we allowed to do this?”, but “How can we move forward fast enough?”
The sun rises beyond the horizon
For PR agencies in Germany, this moment presents a historic opportunity. What has always set good communications consultants apart – developing narratives that stand up to critical scrutiny, cultivating relationships with (editorial) gatekeepers, and crafting messages that resonate – is becoming a key strategic asset in the LLM-driven world.
In this sense, EOS is more than just a model. It is the new guiding principle in the emerging LLM era.