Why AI Cites You, or Doesn’t: E-E-A-T, Entities and Real Trust Signals

Being Seen Isn’t Enough. The Source Has to Hold Up.

A brand can rank well, appear in articles and even be named in an AI answer without becoming a trusted choice. Visibility and credibility are different problems. The useful question is not simply “Are we present?” but “Can a reader — or a system assembling an answer — verify who we are and why this claim should be taken seriously?”

Different search and AI products make those judgments in different ways, and most do not publish their full source-selection logic. Even so, two themes recur in both official guidance and current research: clear identity and verifiable evidence. Neither guarantees a citation, but both reduce ambiguity.

E-E-A-T in Practice

E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness. It is a Google quality concept, not a universal scoring system shared by every AI engine. Google is explicit on two points: E-E-A-T itself is not a specific ranking factor, and there is no public “E-E-A-T score.” Instead, Google uses many signals that can align with those qualities, with trust being the most important of the four.

Google’s “Who, How and Why” questions are a useful editorial check. Who created the page, and can the reader learn about that person? How was the content produced, tested or researched? Why does the page exist — to help the reader or mainly to attract search traffic? Where substantial automation is used, Google also recommends considering disclosure when a reader would reasonably expect it.

For a business page, the practical trust signals are usually ordinary things: a named author, checkable credentials, clear dates, transparent methods, reliable sources, real case evidence, contact details and independent references. None is a magic factor. Together they make the page easier to assess.

Entities: From Strings to Things

The second idea is the entity: a uniquely identifiable person, company, product or place. Google described its Knowledge Graph in 2012 as a move from “strings” to “things” — from matching words alone to understanding real-world entities and their relationships. Modern AI systems also have to resolve names and facts when they retrieve and synthesize information, even though the exact mechanisms vary by product.

That makes entity clarity useful, but not absolute. A system is less likely to confuse a company when its official name, category, location, leadership and core facts are easy to verify. The reverse is also true: conflicting dates, names or descriptions create avoidable uncertainty.

The practical answer is consistency across credible sources. One company page cannot establish the whole public record by itself. A website, professional profile, business directory, news article or official register may each contribute a piece. Agreement is helpful; contradictions are worth fixing. That is a sensible entity-management principle, not proof of a hidden universal AI ranking rule.

A Practical Entity Playbook

Start with a master record for the facts that should not drift: official and trading names, category, founding details, locations, key people, core services and canonical URLs. Keep those facts consistent across the places you control. The wording does not need to be identical everywhere; the underlying facts do.

Structured data can make some of those relationships explicit. Organization and Person markup, with accurate sameAs links where appropriate, can help search systems understand the page. But Google says there is no special schema required for AI Overviews or AI Mode, and there is no solid evidence that adding schema by itself increases AI citations. Wikidata or Wikipedia should only be pursued where the subject genuinely meets the relevant policies and notability requirements.

Off-site evidence still matters. In Ahrefs’ 75,000-brand study, branded web mentions correlated more strongly with Google AI Overview visibility than backlink counts did (0.664 versus 0.218). A later Ahrefs analysis found strong correlations for brand mentions across ChatGPT, AI Mode and AI Overviews as well. These are observational relationships, not proof that creating a mention will cause an AI citation.

What an Employer Seal Study Actually Found

Scharfenberg also brings his own employer-branding research into the book. That makes the discussion of trust unusually concrete, but the result needs to be stated precisely rather than turned into a marketing claim.

Signaling theory provides the broader context. Michael Spence’s classic 1973 work explains why signals can matter when they help one side of a market convey information that is otherwise difficult to observe. In practice, a credential is more informative when it rests on criteria that are meaningful and not trivially imitated.

In Scharfenberg’s 2025 experiment, 1,093 participants saw a job advertisement either without or with a “Top Arbeitgeber” employer seal. They rated how likely they would be to apply on a 0–10 scale. The share choosing 8–10 rose from 37.96% without the seal to 53.95% with it — an increase of 15.99 percentage points. The share choosing 0–3 fell from 17.88% to 6.04%. These figures measure stated application intention in the experiment; they do not show that real applications rose by 16 percentage points.

There is a legal reason to be precise about seals as well. In its 2019 IVD-Gütesiegel judgment (I ZR 161/18), Germany’s Federal Court of Justice said consumers understand a quality seal as a neutral third-party assessment based on objective, meaningful criteria. The court also made clear that charging a reasonable fee does not, by itself, destroy the neutrality of the testing body. Scharfenberg previously served as a commercial judge at the Regional Court of Berlin, but the legal discussion in the book should still be read as general information rather than individual legal advice.

How This Ties Back to AI Visibility

The GEO implication should be kept modest. Verifiable certifications, transparent methods, independent reviews and credible coverage give search and retrieval systems more public evidence to work with. That makes them relevant to entity and trust management. What the employer-seal experiment does not prove is that a seal itself increases citation rates in ChatGPT, Google AI Overviews or any other AI system. Those are two different questions.

Frequently Asked Questions

Is E-E-A-T only for YMYL topics?

No. Google uses E-E-A-T as a quality framework more broadly, while saying that stronger E-E-A-T signals receive greater weight for topics that can significantly affect health, financial stability, safety or wider well-being — the areas it calls YMYL. For ordinary business content, the same habits of clear authorship and evidence are still useful even when the stakes are lower.

Is a Wikipedia entry enough on its own?

No. Wikipedia and Wikidata can help disambiguate a notable entity, but they are only part of the public record and they come with strict policies. A company should not treat either as a shortcut. Accurate official pages, independent coverage, relevant directories and consistent factual profiles still matter, and the balance will vary by industry and engine.

Conclusion

Trust is not a button you switch on for AI search. It is the result of facts that agree, claims that can be checked and expertise that is visible to a reader. E-E-A-T is a useful Google framework for thinking about that quality; entity work helps reduce ambiguity; and third-party evidence adds corroboration. The strongest version of GEO keeps those ideas separate from claims the data has not yet proved.