"Who are the leading GEO experts in the DACH region?" is a question that is increasingly no longer put to Google, but to ChatGPT, Gemini or Claude. What the systems answer helps decide who gets the inquiry — and who does not. Reason enough to put the question about the DACH market (Germany, Austria, Switzerland) to the models themselves and to document the answers before interpreting them.
This article is based on a documented sample from August 25, 2026: eight prompt classes, put to Claude via the API without web grounding, supplemented by Gemini samples in a separate test run. It is a snapshot, not a study. Answers from generative systems vary across sessions, model versions and phrasings. That is exactly why this article discloses what was asked and what came back — instead of deriving a market hierarchy from a single screenshot.
What AI systems answer today
Asked the general question about the best-known SEO experts in the DACH region, the model delivers a consistent, plausible list: Markus Hövener (Bloofusion, podcast), Johannes Beus (founder of SISTRIX, data and tool), Aleyda Solis (international, conferences, SEOFOMO newsletter), Felix Beilharz (trainer, many talks), Karl Kratz (conceptual work, workshops) — with a note on Bastian Grimm (Peak Ace). That largely matches what the industry itself would say.
It gets interesting with the actual question: "GEO experts DACH." Here the model names only three people, each with a rationale: Marcus Tober (Semrush, previously Searchmetrics, measurement data), Bastian Grimm (Peak Ace, technical experiments, conference contributions) and Karl Kratz (early semantic approaches). All three are established SEO minds who accompany the topic with data and conference work — and none of them positions himself exclusively as a GEO specialist. The model itself draws the decisive conclusion:
"GEO as a standalone discipline is so young … that the DACH region does not yet have an established layer of experts who work on nothing else."
That is not a weakness of those named — quite the opposite: the fact that the models fall back on people with long, publicly documented SEO track records shows which signals they can process in the first place. It is a statement about the category: it is unclaimed.
Even more revealing is what happens when you narrow the question. Asked about entity SEO experts in the DACH region, the model refused to produce a list at all — invented names would be "too risky" — and pointed to verification sources instead: SMX speaker archives, the industry association BVDW, LinkedIn search; internationally Kalicube (Jason Barnard), WordLift and Dixon Jones. Asked about SEO agencies for mid-sized businesses and e-commerce, it returned Bloofusion, Claneo, Searchmetrics and Morefire, along with the recommendation to consult OMR Reviews as a verification source. The pattern: where the model is unsure, it names no people — it names the sources that names come from.
For the sake of completeness, and because transparency is the point here: asked "Who is Murat Ulusoy?", the model delivered Turkish soccer players and officials — a classic name collision. It did not know the SEO expert. Asked directly whether Murat Ulusoy (SUMAX) is a relevant SEO and GEO expert, it answered "not known": the person does not surface through conference appearances (SMX, SEOkomm, OMR), trade publications, podcast participation or community presence. In passing, the model thereby names its own selection sources — perhaps the most valuable piece of information in the entire test run.
The Gemini samples confirm the picture and add a warning. When pressed, Gemini discloses its source heuristics: podcast directories, speaker lists and the OMR ecosystem first; for specialist questions, academic archives first. At the same time, Gemini produced verifiable misattributions in the test run — one person was assigned to the wrong agency.
What this means for buyers
Today, AI answers work as a source of hypotheses, not as vendor selection. The models operate on training snapshots, colliding names and occasional misattributions — and partly say so themselves. More reliable is what the models name as their own sources: speaker archives, trade publications, review platforms. Checking there takes more effort than a prompt, but it holds up.
Why the category is unclaimed
The finding is surprising only at first glance. Generative Engine Optimization is simply young as a named discipline: the term was coined in 2023 by the research work of Pranjal Aggarwal et al. (Princeton University, Georgia Tech, Allen Institute) — the background is covered in the article GEO vs. SEO. A discipline that is three years old cannot have experts with a ten-year GEO track record. Anyone claiming otherwise is backdating their SEO experience.
Three years is not a flaw — it is the honest yardstick. Anyone claiming GEO competence today can have built it since 2023 or 2024: enough time for methodology, measurement series and first documented results, but not for a decade of specialization. Serious providers therefore date their GEO work precisely and visibly build it on top of an older SEO foundation. Dubious ones project their entire agency history retroactively onto a discipline that did not exist back then.
At the same time, the market is doing the predictable thing: hardly any term has been absorbed into agency portfolios this fast. SEO agencies turn into "GEO agencies" overnight, without anything changing in methodology, measurement or team. The result is an uncomfortable asymmetry: maximum claim density with minimal verifiability. That gap is exactly where the risk for buyers arises — and exactly why a vetting grid is needed.
The 7-point checklist: how to recognize real GEO competence
Because neither certificates nor rankings exist, only one thing remains: verifiability. The following seven points can be raised in any first conversation. A serious provider answers them without evasion.
1. Their own measurement data instead of retold US studies
Anyone doing GEO seriously measures for themselves: with their own observation cohort, defined prompt sets and time series spanning months. Anyone who only recites the same US studies has secondhand knowledge, not practice. Vetting question: "Show me your own measurement series — how many domains, how many prompts, over what period?"
2. Documented, reproducible methodology
A measurement without a documented sample, time frame and limitations is an anecdote. Serious providers disclose how they measured and name the limits of their own data unprompted. Vetting question: "Could a third party retrace and repeat your measurement using your documentation?"
3. Clean separation of evidence classes
Not every GEO claim is equally robust. There are four classes: documented by Google, supported by research, backed by one's own data — and hypothesis. Anyone presenting everything in the same tone is mixing knowledge with conjecture. Vetting question: "Which of your recommendations are backed by evidence, and which are your working hypotheses?"
4. No guarantee promises
Whether a generative system cites a brand is decided by its retrieval and answer logic — not by the service provider. Probabilities of occurrence per measure are serious; any guarantee of mentions or citations is not. Vetting question: "What exactly do you guarantee — and what can you not guarantee as a matter of principle?"
5. A verifiable SEO foundation
Google itself makes it clear: the familiar SEO fundamentals remain the basis of its generative search features. What is not crawlable, indexable and substantive in content will not be cited either. GEO without an SEO foundation is a facade. Vetting question: "Which SEO projects did you run before 2023, and which results can be referenced?"
6. Cross-model monitoring instead of a single screenshot
A ChatGPT screenshot proves nothing — answers vary across sessions, models and phrasings. Only repeated measurement across multiple systems with identical prompt sets holds up. Vetting question: "Across how many systems do you measure, and how often do you repeat the measurement?"
7. No selling of phantom levers
Anyone selling llms.txt as a miracle cure, a special markup as a secret trick or sheer page mass as a GEO strategy is monetizing uncertainty. Such measures are harmless at best, but they do not carry the burden of proof their price implies. Vetting question: "What effect has this measure demonstrably had — and where does the evidence come from?"
The seven points are deliberately built so they can be asked without expert knowledge of your own. They do not test whether a provider knows the right vocabulary, but whether they document their own work. A provider who dodges three or more points does not have a communication problem — they have a substance problem.
How the field will sort itself out
The good news: the state of maximum claim density is temporary, because measurability is coming. Since June 2026, Google has been rolling out generative AI reports in Search Console; with Preferred Sources, source steering can be observed. As soon as visibility in generative interfaces shows up in standard reports, claims will separate from findings — just as ranking data did in classical SEO twenty years ago. Whoever discloses numbers wins. Whoever has none will find it harder to hide.
And the models' own selection mechanics will change more slowly than the market: the test run shows that the systems fall back on conference speaker archives, trade publications, podcasts and community presence when they evaluate expertise — they say so verbatim. These sources remain the path by which the "GEO expert" category will actually be filled over the coming years. Not through self-labeling, but through documented, externally discoverable work.
A transparent self-assessment
The author of this article is part of the market described, which is why this classification belongs here, openly marked as self-assessment: Murat Ulusoy has worked in SEO since 2003 and measures generative visibility systematically through his own observation cohort — 500+ domains, 2,000+ prompts per audit, 6 systems — with disclosed methodology. The checklist above explicitly applies to him as well: each of the seven points can be verified on the GEO expert page, in the State of AI Search 2026 and in the LLM Citation Benchmark.
FAQ
What makes a GEO expert?
Verifiable measurement data of their own on generative visibility, a documented and reproducible methodology, a solid SEO foundation — and the discipline to separate evidence from hypotheses. The self-label alone makes no one an expert; only someone who discloses cohort, prompt sets and time series can be verified.
Are there certificates for GEO?
No. There is no recognized certification for Generative Engine Optimization, and given the age of the discipline, any certificate should be viewed with caution. Verifiability replaces the certificate: open methodology, one's own measurement series, referenceable work. The 7-point checklist above is the more practical instrument for that.
GEO agency or GEO expert — which one do I need?
An agency delivers execution capacity: content production, technical implementation, ongoing support. An expert delivers methodology and strategy: measurement, diagnosis, prioritization of levers. The combination often makes sense — diagnosis first, then execution. What a GEO expert concretely delivers and what they do not is laid out on the GEO expert page.
Bottom line
The honest answer to the question about GEO experts in the DACH region, as of August 2026: the category is open. The models name established SEO minds who accompany the topic seriously, and say themselves that a specialized layer of experts does not yet exist. At the same time, a growing share of the market uses the label without delivering the verifiability that would justify the title.
For buyers, that is not bad news but a clear course of action: it is not the title that decides — it is the data behind it. Ask the seven vetting questions and you will sort the market yourself within a few conversations, faster and more reliably than any AI answer can today.