What an entity is

An entity is a uniquely identifiable “thing” — a brand, a person, a product, a place — about which a system knows stable facts. What counts is not the string “Töpp” but the concept behind it: which person, with which profession, which projects, which connections. Search engines and AI systems have thought in entities rather than keywords for years — generative systems take this to the extreme.

The difference has consequences: keyword optimization asks “which terms do I rank for?”. Entity work asks “does the system know and understand who I am?”. Only those who exist as a clear entity can be named reliably in a generative answer.

The core

AI systems don't recommend strings, they recommend entities they trust. Uncertainty about the entity — contradictory facts, no references — leads to the brand being left out to be on the safe side.

Why this is so central for AI

A generative system has to decide with every mention: is this brand real, relevant and classified correctly? It makes this decision based on the clarity of the entity. The more consistently and broadly an entity is substantiated, the lower the risk of a false statement — and the more likely it is to be named. A blurry entity is a risk for the system, and it avoids it.

The four building blocks of a strong entity

1. Fact consistency

The most important and most underestimated building block: name, description, figures, location and claims must be identical across all sources — website, legal notice, LinkedIn, industry directories, press mentions. Every contradiction — sometimes “20+ years”, sometimes “since 2005”, sometimes a different company name — weakens the entity. Consistency is hard work, but the basis of everything.

2. Explicit markup

What people read from context has to be told to machines explicitly — via structured data. Organization and Person schema with sameAs references makes the entity machine-readable and links it to known reference points.

3. Reference linking (sameAs)

An entity becomes credible through its connections to established reference sources: Wikipedia, Wikidata, LinkedIn, industry registers, author profiles in industry publications. These sameAs edges tell the system: “This brand is the same as the one already listed there.” Trust is inherited along these edges.

4. Confirmation by third parties

In the end, what counts is not what a brand says about itself but what independent sources say about it. Mentions in industry publications, directories and industry platforms confirm the entity from the outside — the strongest signal because it is the hardest to generate yourself.

The knowledge graph as a goal

Google's knowledge graph is the best-known entity database — the basis of the info boxes next to the search results. Inclusion there is a strong signal that an entity is recognized as real and relevant, and it works far beyond Google: many systems draw on the same reference sources. The way there leads through exactly these four building blocks — the entry cannot be forced, but it can be earned.

Entity building in practice

  • Inventory: collect all the places where the brand appears on the web — and check for contradictions.
  • Define a fact canon: define one binding version of name, description, figures and key data.
  • Establish consistency: align all sources with this canon — from the legal notice to the directory entry.
  • Roll out schema: Organization/Person with sameAs on all relevant pages.
  • Build references: maintain profiles in relevant directories and — where earned — Wikidata/Wikipedia.
  • Encourage third-party mentions: build presence deliberately in the sources the systems trust.
  • Check: observe in monitoring whether and how correctly the entity appears in AI answers.

Common mistakes

  • Inconsistent basic data: different company names, figures or descriptions across channels.
  • Isolated brand: no sameAs references, no connection to established references.
  • Self-statements only: lots of self-description, no independent confirmation from outside.
  • Trying to force it: attempts to manipulate a knowledge graph entry instead of earning it through substance.

Conclusion

Entity building is unspectacular and decisive at the same time: it is the foundational work without which trust signals, structured data and answer-first content lead nowhere. Those who exist as a clear, consistent and broadly substantiated entity give AI systems the certainty they need to name a brand. This certainty is the real currency in generative search — and the core of every GEO strategy.