Entity-Based SEO: Optimizing for What Search Engines Actually Understand
Search engines stopped reading queries as strings around 2013. Most SEO content has not caught up. Entity-based SEO is the practice of writing for the entity layer that retrieval systems actually use.
The Hummingbird update in 2013 was the moment Google started reading queries as references to entities rather than as keyword strings to match. Twelve years later, most SEO content still teaches keyword density and exact match as if neither shift happened. Entity-based SEO is the practice that catches up.
What an entity is, precisely
An entity is a concept that exists in the world independently of any specific string that names it. Topical authority the concept, the word topical-authority, and the phrase SEO authority on a topic all refer to the same entity. Three strings, one entity.
Search engines and language models maintain internal representations of entities and the relationships between them. Google’s Knowledge Graph is one implementation; language models build implicit entity representations from training data. The entity layer sits underneath the string layer and is what determines whether your content gets surfaced for queries about the entity.
Entities vs keywords: what changes
Question
Keyword-based SEO
Entity-based SEO
Unit
String
Concept
Optimization target
Match exact-phrase queries
Resolve to the correct entity
What a page is
A response to one keyword
The canonical reference for one entity
Internal linking
Anchor-text density on keywords
Topic-coherent links between entity pages
Failure mode
Keyword cannibalization
Definitional drift across entity pages
What ranks better
Pages with the right density
Pages embedded in the right entity cluster
AI citation
Inconsistent
Strong, because entity consistency is the citation signal
How entity resolution works
When a user types topical authority, Google’s system does three things roughly in parallel. It matches the query string against indexed documents (lexical matching). It encodes the query as a vector and finds documents with similar embeddings (semantic matching). And it resolves the query to an entity in its knowledge graph and finds documents that map to the same entity (entity-based retrieval).
A page with strong keyword density on the literal string may rank on lexical matching while losing on entity resolution if the page covers topical authority in a way that does not map cleanly to the entity. A page that names the entity once and covers it canonically may rank lower on density while ranking higher overall, because all three layers reinforce.
Three things entity-based SEO actually requires
One canonical page per entity. The page is the site’s answer to “what is this entity” and to “how do you use it.” Multiple pages on the same entity produce ambiguity that retrieval systems resolve in unpredictable ways.
Consistent definitions across the network. Every page that references the entity uses the same definition, lifted verbatim if necessary. The site reads as a single source on the entity rather than as a committee of sources that disagree.
Topic-coherent internal linking. Pages on related entities link to each other. Pages on unrelated entities do not. The link-graph topology should approximate the entity-graph topology of the topic.
The Knowledge Graph connection
Google’s Knowledge Graph stores entities and their relationships explicitly. Some entities have knowledge-panel coverage (the boxes on the right of branded queries); most do not. Whether your entity has explicit Knowledge Graph representation is less important than whether your content maps cleanly to whatever representation exists.
Structured data (schema.org markup) is the mechanism that ties your content to known entities explicitly. The sameAs property links your representation of an entity to its canonical URL elsewhere (Wikidata, Wikipedia, your own canonical page). The link tells search engines: these references point to the same entity.
The practice of optimizing content to map cleanly to the entities (concepts in the world) that retrieval systems use, rather than to the keyword strings that name them. The unit is the entity; optimization means coverage, consistency, and topic-coherent linking around each entity.
What is an entity in SEO?
A concept in the world that queries refer to, independent of any specific string. The fruit apple and the company Apple are two different entities the string “apple” can refer to.
How is entity-based SEO different from semantic SEO?
Entity-based SEO emphasizes the entity-resolution side: making content map cleanly to the right entities. Semantic SEO is the broader practice that also covers the graph layer and the context layer. See the pillar.
Do I need schema markup for entity-based SEO?
Schema markup helps. It is not strictly required. The underlying entity coverage and consistency matter more; markup is the affirmation, not the substance.
What is the difference between an entity and a keyword?
An entity is a concept; a keyword is a string. One entity can have many keywords that refer to it. Optimization for the entity covers all the strings that refer to it; optimization for one keyword misses the others.