AEO Agencies in Germany Tested: 12 Agencies, 8 Checks on Their Own Websites

Dominik Breitbach Founder and Lead SEO Strategist at taismo

Dominik Breitbach founded taismo GmbH and works as lead SEO strategist on SEO retainers and visibility inside AI answers. He runs taismo as an SEO and GEO agency from Munich and advises companies in German and English.

⏱ Reading time: 14 min🔄 Last updated: 11 October 2026

We tested 12 AEO agencies in Germany on October 4, 2026 to see whether they apply answer engine optimization to their own websites. AEO, also called generative engine optimization or GEO, makes a company citable for ChatGPT, Gemini, Perplexity, and Google’s AI features. The test uses 8 checks that anyone can verify from the outside: Wikidata, ISNI, sameAs profiles, structured data beyond a plugin’s default, a Person node, speakable, llms.txt, and consistent NAP data. By our measurement, taismo meets 7 of the 8 checks, HECHT INS GEFECHT and SUMAX meet 5 each, and every other agency meets 4 or fewer. taismo publishes this test, so every check comes with the method you can use to verify the result yourself.

👉 Want to see how your own website performs on these checks? Run the free AI visibility check

Transparency: This test was created by taismo, an SEO and GEO agency from Munich, and taismo ranks first in it. We chose checks that anyone can verify without special tools. All values refer to the cutoff date of October 4, 2026.

The results: 12 AEO agencies, 8 checks, one cutoff date

The table shows how many of the 8 checks each agency meets on its home page and in public registries. When two agencies meet the same number of checks, the number of structured data values on the home page decides the order.

Three findings stand out. speakable and an ISNI entry appear only at taismo. With 650 structured data values, the taismo home page carries the most extensive JSON-LD graph in the test. And the spread in structured data is wide: It ranges from 0 to 650 values on the home page.

Why we test the agencies’ own websites

An AEO agency sells visibility in AI answers. For ChatGPT, Gemini, Perplexity, or Google’s AI Overviews to name a company as a source, that company has to be unambiguous for machines. That clarity comes from building blocks that almost every AEO agency describes on its service page: The entity of the brand, structured data, citable content, consistent company data, and entries in knowledge bases.

The agency’s own website is the most reliable proof of whether it masters these building blocks. It is the only project with no client approval, no outside budget, and no competing vendor in the way. Any gap on the agency’s own website is a gap the agency controls itself.

For international companies, this test solves a second problem. If you are a US or UK business looking for an AEO partner in Germany, you usually cannot judge German-language case studies, German reviews, or German award listings. The 8 checks work without reading German, because they look at code, registries, and data that are the same in every language.

Existing agency lists mostly rate reviews, team size, references, or prices. In the 32 German agency lists we reviewed on October 2, 2026, we found none that measures whether the listed agencies apply their service to their own websites. This test closes that gap. For a ranked list with agency profiles, reviews, and client voices, see our overview of the best GEO agencies in Germany.

Method: How we measured the 12 AEO agencies

The test measures 8 yes or no checks. Each check describes an AEO building block that the agency has either implemented on its own website or in a public registry, or has not.

1. Wikidata: The agency has its own Wikidata item as an organization.
2. ISNI: The agency is listed as an organization in the international ISNI registry.
3. sameAs: The organization node in the home page JSON-LD links to at least one external profile via sameAs.
4. Schema above the plugin default: The home page JSON-LD contains more than 100 values.
5. Person node: The home page JSON-LD contains at least one person, typically the managing director.
6. speakable: The home page JSON-LD marks the key statements of the page with speakable.
7. llms.txt: The URL domain/llms.txt returns a text file with status 200, not an HTML error page.
8. Consistent NAP: Name, address, and phone number in the legal notice match the Google Business Profile.

Five rules apply to the measurement:

• Cutoff date: All values date from October 4, 2026. Websites change, and every value applies to that day only.
• Where we measured: We measured each agency’s home page in the delivered HTML without JavaScript, with the user agent of a regular browser, one domain after another.
• Structured data values: We counted every value in every JSON-LD block on the home page, so every type, address, name, and URL counts as one value.
• Threshold for check 4: The default output of Yoast SEO came to between 49 and 55 values on the three home pages that use Yoast alone. More than 100 values therefore means a website is well above what a plugin generates by itself.
• Ties: With an equal number of checks met, the number of structured data values decides.

The selection: We measured 12 agencies that explicitly position themselves as providers of AEO or GEO (the term German agencies mostly use) and have a dedicated service page for it. The field includes the German agencies that AI answers mention most often when asked about such agencies, agencies from German best-of lists, and taismo itself. The home pages measured were taismo.de, hechtinsgefecht.de, sumax.de, evergreen.media, seowerk.de, gerlach.media, suchhelden.de, hurra.com/de, geoagentur.de, aufgesang.de, claneo.com/de, and dreikon.de.

The full measurement matrix: All 8 checks for all 12 agencies

✓ = met, ○ = not found on October 4, 2026. The number in the sameAs column counts the unique profile URLs on the organization node of the home page.

¹ seowerk: The Augsburg profile matches the legal notice. The Munich profile lists a phone number that does not appear in the legal notice. We counted the check as met in the agency’s favor. ² hurra.com: The delivered HTML contains no JSON-LD. Markup that a tag manager injects with JavaScript is outside the scope of this measurement. ³ geoagentur.de: We found no Google Business Profile under the brand name.

The matrix shows a clear pattern. Checks that a plugin or web designer handles along the way are common: 11 of 12 agencies have consistent NAP data, 7 of 12 have an llms.txt. Checks that require deliberate work on the agency’s own entity are rare: A Person node at 5 agencies, speakable and ISNI at one, Wikidata at none.

The 8 checks in detail: Result and how to verify it

For each check, you will find what it says about AEO, how the 12 agencies performed, and how to verify the value yourself. Each verification takes a browser and a few minutes.

1. Wikidata: The most demanding check is open at all 12 agencies

Wikidata is the free knowledge base of the Wikimedia Foundation. A Wikidata item describes a company with a unique identifier, founding date, headquarters, and website, and links it to other items. Many knowledge graphs match identities against Wikidata. What an entry does for a company and which rules apply there is covered in our glossary entry on Wikidata.

Result: We found no Wikidata item as an organization for any of the 12 agencies on October 4, 2026. Wikidata requires evidence from independent sources and deletes entries that only serve self-promotion, which makes this check the most demanding of the 8.

How to verify: Enter the company name in the search on wikidata.org. For a more thorough check, query whether any Wikidata item lists the agency’s domain as its website: https://www.wikidata.org/w/api.php?action=query&list=exturlusage&euquery=agency.com, with agency.com replaced by the domain.

2. ISNI: Only taismo is registered as an organization

The ISNI (International Standard Name Identifier) is an ISO standard identifier for persons and organizations involved in creating or publishing works. Libraries, publishers, and databases use it to resolve name ambiguity. For a brand entity, the ISNI is an identity record from a registry the agency does not manage itself.

Result: Since September 21, 2026, taismo has been listed as an organization under ISNI 0000 0005 3161 3181. For the other 11 agencies, the registry search on October 4, 2026 returned no match.

How to verify: Search the company name on isni.org. The site protects itself with a browser check and answers automated requests with status 403. To check without a browser, use the public search interface https://isni.oclc.org/sru/?operation=searchRetrieve&recordSchema=isni-b&query=pica.nw+%3D+%22CompanyName%22; the value under numberOfRecords is the number of matches.

3. sameAs: 6 of 12 agencies link external profiles

The sameAs property connects the organization node in the JSON-LD with profiles of the same organization on other platforms, such as LinkedIn, review portals, or business directories. It tells a machine which profiles on the web belong to this company. For the Knowledge Graph and for AI systems, that is a direct hint for matching.

Result: 6 of 12 agencies use sameAs on the organization node: HECHT INS GEFECHT (36 profiles), taismo (32), SUMAX (6), Evergreen Media and seowerk (4 each), and Gerlach Media (2).

How to verify: Open the home page source with Ctrl+U (Cmd+Option+U on a Mac) and search for sameAs. Make sure the list belongs to a node of type Organization or LocalBusiness. Count identical profiles in two spellings only once.

4. Structured data: 4 agencies go beyond the plugin default

Structured data describes a page in a fixed vocabulary that search engines and AI systems read without interpretation. An SEO plugin generates a basic set of website, web page, organization, logo, and breadcrumbs. Everything beyond that is handwork: Services, people, awards, locations, or defined terms.

Result: 4 agencies are above the threshold of 100 values. taismo leads with 650, followed by Evergreen Media with 212, HECHT INS GEFECHT with 161, and SUMAX with 138. Aufgesang (55), Claneo (53), and seowerk (49) deliver the default output of Yoast SEO. DREIKON delivers 12 values, hurra.com none in the delivered HTML.

How to verify: Enter the home page URL in the Schema Markup Validator at validator.schema.org. It lists every node with its properties. To get the number of values, count them or paste the JSON-LD from the page source into any JSON tool. How we build such graphs for clients is shown in our guide to schema markup in practice.

5. Person node: 5 agencies show the people behind the brand

A Person node describes a human being with name, role, and profiles, and links that person to the organization. For AI systems, it makes visible who stands behind a brand and what that person is known for professionally. At agencies, that is usually the management.

Result: 5 agencies list people in the home page JSON-LD, namely taismo, Evergreen Media, HECHT INS GEFECHT, SUMAX, and Gerlach Media.

How to verify: Search the page source for “@type”:”Person” or read the node list in the Schema Markup Validator. Only people in the JSON-LD of the home page count.

6. speakable: Only taismo marks its key statements

The speakable property points to the passages of a page that are suited for being read aloud, using CSS selectors. Google describes it as a beta feature for voice assistants. For AEO, speakable signals which sentences of a page carry the key statement. It only works if the selectors match a real element on the page.

Result: On October 4, 2026, we found speakable markup only on the taismo home page.

How to verify: Search the page source for speakable. If it contains a cssSelector such as .lead-def, search the source for exactly that class. The markup is effective only when both are present.

7. llms.txt: 7 of 12 agencies provide one

The llms.txt is a text file in the root directory of a website. It summarizes in Markdown who the company is and which pages carry the most important content, written for language models. Jeremy Howard (Answer.AI) proposed the format in September 2024, and it is not a binding standard yet.

Result: 7 agencies serve an llms.txt, namely taismo, HECHT INS GEFECHT, SUMAX, seowerk, Suchhelden, hurra.com, and geoagentur.de. At Evergreen Media, Gerlach Media, Aufgesang, Claneo, and DREIKON, the URL returns status 404.

How to verify: Open agencydomain.com/llms.txt in your browser. A real llms.txt appears as plain text with headings and links. If the browser shows a designed error page or the home page, the file is missing.

8. NAP data: 11 of 12 agencies are consistent

NAP stands for name, address, phone. When these details match across the website, the Google Business Profile, and directories, the company can be identified without doubt. Contradictions make search engines and AI systems doubt a source. Our glossary entry on NAP explains the details.

Result: At 10 agencies, the legal notice and the Google Business Profile match completely. At seowerk, the Augsburg profile matches, while the Munich profile lists a phone number that does not appear in the legal notice. For geoagentur.de, we found no Google Business Profile under the brand name.

How to verify: Search the company name with its city on Google Maps and compare name, address, and phone number with the legal notice (in Germany called the Impressum, legally required on every business website). If there are several locations, check each profile separately.

What the test covers and which questions need further sources

The test measures diligence on the agency’s own website. It shows whether an agency applies the AEO building blocks it sells to itself. Agencies with many checks met have proven that they master the technical side of AEO and invest time in their own entity.

Four questions need additional sources:

• Quality of client work: An agency with few checks met can still deliver excellent results for clients. Ask for references with verifiable numbers in the first call.
• Actual mentions in AI answers: How often ChatGPT or Gemini name an agency also depends on mentions on third-party sites, reviews, and trade publications. How to measure such mentions is covered in our guide to AI search visibility.
• Markup loaded later: JSON-LD that a tag manager injects with JavaScript is invisible to a measurement of the delivered HTML. Google’s Rich Results Test executes JavaScript and finds such markup.
• Subpages: We measured the home page. Some agencies mark up their service or team pages in more detail.

Beyond the count, the connections matter: Good structured data matches the visible content and links its nodes into one graph via @id. The Schema Markup Validator shows you both.

How taismo implements the 8 checks

At taismo, AEO is a fixed part of every SEO retainer. We build a company’s entity, connect it through structured data, write citable content, and measure mentions in the major AI systems. What that looks like for your company is described on our page about AEO and GEO services for AI visibility.

On our own website, the cutoff date of October 4, 2026 shows: A connected JSON-LD graph with 650 values on the home page, a Person node for managing director Dominik Breitbach, speakable markup, 32 linked profiles, a bilingual llms.txt in German and English, and ISNI 0000 0005 3161 3181. The company data on the website and in the Google Business Profile match. We work with clients in German and English, which matters if you run your AEO for the German market from abroad.

How to evaluate an AEO agency: 8 checks plus 3 questions

The 8 checks work as a pre-selection. They cost you a few minutes per agency and show who practices AEO. For the final decision, three questions complete the picture:

1. References: Which clients does the agency name, and which results can you verify for those clients?
2. Measurement: How does the agency measure AI mentions, with which set of prompts, and for which AI systems?
3. Integration: Is AEO part of an ongoing SEO retainer, or a one-off project with no maintenance afterward?

If you hire from outside Germany, add a fourth question: In which languages does the agency deliver content, reporting, and calls? AEO for the German market needs German content that answers German prompts, while you need reporting you can read.

An agency that performs well on the 8 checks and answers these questions convincingly is a solid choice. Feel free to measure us by the same standards.

Frequently asked questions about AEO agencies in Germany

What is an AEO agency?

An AEO agency optimizes a company’s website and brand so that answer engines such as ChatGPT, Gemini, Perplexity, and Google’s AI Overviews name and cite it. The work covers citable content, structured data, entity building, and the measurement of AI mentions.

Is there a difference between AEO and GEO?

AEO (answer engine optimization) and GEO (generative engine optimization) describe the same goal: Being named in AI-generated answers. AEO is the more common term in the US market, GEO dominates in Germany, and the methods overlap almost completely.

How do I evaluate an AEO agency?

Start with the agency’s own website: Connected JSON-LD with Organization, Person, and sameAs profiles, a reachable llms.txt, and consistent company data can be checked in a few minutes. Then ask for references with verifiable results and for the prompt set the agency uses to measure AI mentions.

Does every website need an llms.txt?

The llms.txt file is a proposal and not a binding standard. It takes little effort and summarizes your most important pages for language models, which makes it worthwhile above all for companies with services that need explaining.

What does speakable do in structured data?

The speakable property marks the passages of a page that are suited for being read aloud, using CSS selectors. Google lists it as a beta feature for voice assistants, and it only works if the selectors match real elements on the page.

Does a large schema graph say anything about agency quality?

A large, connected schema graph shows effort and technical skill on the agency’s own website. The quality of client work is decided by references with verifiable results and by monitoring AI mentions over several weeks.

Sources

• Wikimedia Foundation: “Wikidata:Notability”, Wikidata, accessed October 4, 2026.
• ISNI International Agency: “What is ISNI”, isni.org, accessed October 4, 2026.
• Schema.org: “sameAs”, schema.org, accessed October 4, 2026.
• Google Search Central: “Speakable (Article, WebPage) structured data (BETA)”, developers.google.com, accessed October 4, 2026.
• Howard, J. (Answer.AI): “The /llms.txt file”, llmstxt.org, September 3, 2024.
• Schema.org: “Schema Markup Validator”, schema.org, accessed October 4, 2026.