SEOlogie
DE

Gatekeepers / AI & Search Engines — Who Decides Whether You Get Found?

The Five Building Blocks of Being Found · Building block: Gatekeepers / AI & Search Engines · by Ortwin Oberhauser · Last updated: June 2026

Gatekeepers are the systems that stand between a source and the seeker and decide what gets through — and what doesn't. A gatekeeper can be a search engine, an AI system, a platform algorithm. It filters, evaluates and decides: which source is reliable enough to be shown? Which answer is relevant enough to reach the seeker?

You don't persuade a gatekeeper. You're recognized by it — or you're not.

The gatekeeper is the building block in the SEOlogie communication model that stands between everything a source has built — and the human being it wants to reach.

A source can be clearly described. Its answers can be precisely worded. Its presence can be built in the places that matter. And still, all of it can fail at the gatekeeper — if the gatekeeper doesn't recognize the source, doesn't trust it, doesn't let it through.

The gatekeeper is the system that stands between the source and the seeker and decides what gets through.

The seeker's personal assistant

Picture a corporate chief executive who does nothing himself. He reads no emails. Opens no post. Makes no calls of his own. Everything runs through his personal assistant. The assistant filters, summarizes, prioritizes — and decides what reaches the boss at all.

When the boss says, "Get me some bread rolls," the assistant buys them. He decides which bakery. He knows the boss's preferences and brings the rolls that fit them best. If the favourite kind happens to be sold out, he brings the closest match. And if he comes back with the wrong ones, he hears: "Three years, and you still don't know what I like?"

That is exactly how an AI gatekeeper works.

The seeker no longer asks ten different sources himself. He asks one question — of ChatGPT, of Google, of his voice assistant. The AI assistant takes over. It searches, filters, evaluates and brings the seeker the answer that best fits his question. Not ten results. One answer. One source — the one it considers most reliable.

A source this assistant can't recognize never makes it to the boss. A source it can recognize but that doesn't fit its criteria never makes it there either.

Communication no longer happens directly. It goes through the recipient's assistant.

The gatekeeper's rules

Alongside the assistant principle, there's a second image that helps: the bouncer outside a club. He has a list of precise rules. Wearing sandals — you're not getting in. Smelling of alcohol — not getting in. No jacket — not getting in. He passes no judgement on whether someone is a good person or seems likeable. The rule decides.

An AI gatekeeper works much the same way. It has criteria too. And it doesn't deviate from them.

No full name — not verifiable — doesn't get through. Contradictory information across different platforms — not consistent — doesn't get through. Claims with no provable basis — not trustworthy — don't get through.

The decisive difference from the human bouncer: an AI gatekeeper doesn't judge by impressions. It can't assess a photo. It feels no sympathy. It isn't moved by a compelling company story. Looks, brand colours, a beautiful brand film — none of that counts. What counts is what can be verified.

Much of what works in classic corporate communication — a polished appearance, strong design, a likeable presence — has no effect on the AI gatekeeper. What does have an effect is what is structured, consistent and confirmed by other sources.

Two kinds of gatekeepers

External gatekeepers

External gatekeepers stand above the individual platforms. They search the entire web and decide what shows up in answers, search results and recommendations.

Search engines (Google, Bing, Naver, Baidu) — they crawl, index and rank. Whatever isn't indexed doesn't exist for these gatekeepers.

AI answer engines (ChatGPT, Perplexity, Google AI Overviews, Claude) — they synthesize answers from whatever they identify as reliable sources. The answer no longer contains ten blue links — it contains one answer, with one or a few cited sources. Often just one. The AI crowns a single winner per query. There is no second place. If you aren't cited, you're invisible. The age of zero-click has begun: the seeker gets his answer without ever visiting a website.

Voice assistants and AI agents (Siri, Alexa, Copilot, autonomous shopping agents) — they act on the seeker's behalf. A source that isn't recognized doesn't get the job either.

Platform-internal gatekeepers

Every platform has its own internal gatekeeper — the algorithm that decides who gets shown a piece of content. The YouTube algorithm decides which videos get recommended. The LinkedIn algorithm decides which posts get reach. The TikTok algorithm decides what appears on the For You page.

These internal gatekeepers are independent of one another — each with its own criteria, its own signals, its own logic.

What does the gatekeeper look for?

That's the decisive question. And it has two answers — one for classic search engines, one for AI systems.

Classic search engines: ranking factors

Search engines like Google have developed a complex system of evaluation signals over decades. The most important categories:

Technical foundation. Does the website load fast? Does it work on mobile devices? Is it secure (HTTPS)? Can it be crawled and indexed? Without this foundation, nothing else comes into play.

Content. Does the content genuinely answer the seeker's question? Is it clearly structured — with unambiguous headings, understandable language, statements you can follow? Is it up to date?

Authority. Who points to this source? Backlinks from reliable, topically relevant websites tell the gatekeeper: others trust this source. That is one of the strongest signals of all.

Structure. Is the content marked up so machines can read it? Schema.org markup explains to the gatekeeper, in a language it can process directly: What is this page? Who wrote it? What is it about? Which company stands behind it?

AI systems: entity recognition and verification

AI answer engines work differently from classic search engines. They don't look for the best-ranking document — they look for the most reliable source for an answer. The question an AI gatekeeper asks isn't: which website has the most backlinks? It's: which source can I responsibly cite?

The foundation of that decision is entity recognition — the AI's ability to identify a source as a known, real entity.

An entity is a clearly defined, unambiguously identifiable unit in the world: a company, a person, a product, a place, a concept. The AI gatekeeper checks: is this entity known in its data sources? Does it appear consistently — on its own website, on external platforms, in reputable directories, in trade publications? Does the information match up?

Anyone who isn't recognized as an entity doesn't exist for the AI gatekeeper — no matter how good the content is.

People, companies, products — everything must be verifiable

Here the difference between human and machine judgement shows most clearly.

People

A human reads on a website: "Our engineer Thomas H. has 20 years of experience in plant engineering." And thinks: sounds good, seems credible.

An AI gatekeeper reads the same sentence — and can do nothing with it. Thomas H. is not an entity. There's no way to check who Thomas H. is, whether the claim is true, whether he actually has the experience stated.

The same goes for: "Dr. Hannes K. has reviewed and approved the procedure." Which Dr. Hannes K.? Where did he study? In which field? Are there publications, certifications, external references that confirm this person as a real entity?

For AI gatekeepers, people need a digital footprint that can be verified:

Companies

A company as an entity must be recognizable and consistently described:

Products and services

"The product withstands the highest loads." — For an AI gatekeeper, that statement is worthless. It can't be verified.

What can be verified:

Superlatives without evidence — "best quality", "highest precision", "leading provider" — don't signal strength to the AI gatekeeper. They signal the opposite: a source that asserts instead of proving.

The YMYL line

For certain subject areas, an even stricter standard applies. Google has gathered these areas under the term YMYL — Your Money or Your Life.

It means every area in which wrong or misleading information can do people serious harm:

In these areas, good content and a structured profile aren't enough. The gatekeeper demands demonstrable professional authority: licensed professions, academic qualifications, memberships in professional bodies, external confirmation by recognized institutions.

A law firm writing about employment law must be recognizably run by admitted lawyers — not by "a team of experienced experts". A doctor publishing health content needs verifiable proof of a licence to practise. A financial adviser needs their registration number.

YMYL is not a special case — it's the general principle sharpened to a point: claims need evidence. The greater the potential harm from wrong information, the more evidence it takes.

E-E-A-T: the framework behind the judgement

Google has summed up the criteria by which AI gatekeepers evaluate sources under the term E-E-A-T. The letters stand for:

Experience — first-hand experience. Has the person or company actually lived what they write about? A hotelier writing about guest care brings his own experience. An editor covering the same subject without ever having run a hotel doesn't. AI gatekeepers prefer sources with a genuine, traceable background of experience.

Expertise — provable specialist knowledge. Training, qualifications, specialization — documented, named, linked to real evidence.

Authoritativeness — recognition by others. Is this source cited, mentioned, linked by other reliable sources? Authority is not a self-declaration — it arises through external confirmation.

Trustworthiness — reliability and transparency. Do the statements match what other sources say? Have mistakes been corrected openly, with a visible record? Is the source consistent — today and yesterday and a year ago?

E-E-A-T is not a ranking system you can optimize. It's a framework describing what a gatekeeper fundamentally evaluates. If you read E-E-A-T not as a checklist but as a question of stance, you have the right angle: am I actually what I claim to be — and can I prove it?

And the gatekeeper doesn't only evaluate what a source says about itself. Increasingly, it evaluates whether the world tells the same story. Your own statements are checked against what reviews, trade publications, press articles and external mentions report. Trust doesn't arise from self-presentation — it arises from agreement.

Who controls the narrative

Anyone who isn't clearly recognizable as a source hands control of the narrative to the gatekeeper.

And the gatekeeper fills every gap.

If a source doesn't describe itself, others describe it. If it provides no answers, other people's answers get drawn in. If it tells no story, a story arises anyway — out of reviews, comments, press articles, forums. That story can differ sharply from the company's own image of itself.

So the question is not whether a source gets described. The question is: by whom.

If you build your own source to be clear, structured and findable, you keep control over how you're described. That's not a technical problem. It's a strategic decision.

What this means for a source

There is no trick for getting past the gatekeeper. No shortcut. No manipulation that works for long — AI systems keep getting better at spotting manipulation.

What works is the opposite: being a clear, provable, reliable source.

In practice, that means:

You don't persuade a gatekeeper. You're recognized by it — or you're not.

The gatekeepers of tomorrow

What begins today as a recommendation becomes a decision tomorrow.

Today, an AI gatekeeper recommends sources. It shows the seeker which answers, which companies, which products are relevant. The seeker then decides for himself.

Tomorrow, the assistant decides.

AI agents that act autonomously on the user's behalf already exist. They research, compare and buy — with a budget and a credit card, without the user confirming every single step. If you need a hotel, you tell the assistant. It books. If you're looking for a lawyer, you delegate the search. The assistant chooses. If you need a supplier, you let the assistant send the enquiry.

In that world, whether a source is recognized by the gatekeeper is no longer a question of visibility. It's a question of existing in the market at all.

Visibility isn't becoming less important. It's becoming more important than ever before — and it's no longer decided in the moment of the search, but long before: in the moment a source decides whether it wants to be verifiable, structured and recognizable — or not.

Anyone who gets through the gatekeeper reaches the building block everything was built for: The seekers / those who fit.

Sources and literature

Cite this entry

Oberhauser, Ortwin (2026): "Gatekeepers / AI & Search Engines — Who Decides Whether You Get Found?" — SEOlogie, the wiki of the study of letting yourself be found. seologie.com/en/gatekeeper.