Entity-based SEO identifies content gaps by mapping how Google’s Natural Language API recognizes “things” rather than “strings,” as Ray Martinez of Archer Education demonstrates with a 58 recognition score. High coverage scores can be a “lie” if the content lacks clean noun phrases that the machine can ground in the Knowledge Graph. In-house SEO managers and agency content strategists should act on this to improve LLM visibility.
Who should act on Entity-based SEO and who should skip it
Readers who should act on Entity-based SEO
- In-house SEO Managers should act on Entity-based SEO to ensure their brand’s core services are understood by AI search engines.
- SEO Agency Content Strategists should act on Entity-based SEO to provide clients with data-backed “blue ocean” content opportunities.
- Technical SEO Specialists should act on Entity-based SEO to improve schema depth using N-quads and JSON-LD.
- Higher Education Marketers should act on Entity-based SEO to ensure niche industry terms like “stealth applicant” are correctly mapped.
Readers who should skip Entity-based SEO
- PPC Specialists should skip Entity-based SEO and look at Google Ads certification or PPC-specific guides instead.
- Small business owners without technical resources should skip Entity-based SEO and focus on general content marketing best practices.
What to do next about Entity-based SEO
- Extract the site’s existing JSON-LD and HTML files to create a baseline corpus for analysis.
- Pass the corpus through the Google Natural Language API to obtain match status and entity recognition scores.
- Identify entities with “Partial” or “Unrecognized” status to pinpoint specific semantic gaps.
- Define a custom lexicon for niche terms like “stealth applicant” to provide the machine with necessary context.
- Compare the resulting entity map against a competitor crawl to identify “blue ocean” content clusters.
The Shift from Strings to Things
Entity-based SEO moves beyond keyword-based SEO by prioritizing the underlying concepts Google understands. In 2012, Google announced its “things, not strings” shift, signaling a transition toward a system that connects entities back to a Knowledge Graph. This move established the foundation for schema.org and modern information retrieval.
Strings refer to isolated text sequences, such as “OPM,” which can be ambiguous to a machine without context. Things represent the actual entities—like “online program management company”—that Google identifies via unique Machine IDs. Moving to entity-based logic allows for better visibility in AI-driven search results because it aligns content with the way Google organizes information.
Semantic triples and N-quads provide the foundational logic for connecting entities
Semantic triples form the foundational logic for how entities are connected in a knowledge graph by linking a subject, a predicate, and an object. For example, Archer Education (subject) offers (predicate) enrollment marketing (object) creates a basic understanding for the machine. These triples are the core component of the JSON-LD vocabulary used to define brand relationships.
What is a Semantic Triple?
A semantic triple is a set of three connected entities consisting of a subject, a predicate, and an object. It provides the basic logic for how a Knowledge Graph connects different concepts, such as a brand and the service it provides.
N-quads represent the next level of depth by stacking a fourth entity onto a semantic triple to provide specific context. By adding a fourth node, an SEO can clarify that “Archer offers enrollment marketing through a GEO audit,” ensuring the answer isn’t vague. Layering these entities prevents the machine from making incorrect assumptions about a brand’s specific niche.
What is an N-quad?
An N-quad is a semantic structure that adds a fourth entity to a standard triple. This extra layer provides more specific context, allowing a machine to distinguish between similar services or business models.
The Scoring Framework: Recognized vs. Partial vs. Unrecognized
The Google Natural Language API provides a scoring framework that categorizes every entity in a corpus into three buckets. Recognized entities have a Machine ID and are understood by Google, while Unrecognized entities are missing from the site entirely. Partial recognition occurs when vague language prevents the machine from identifying the specific entity intended by the author.
Ray Martinez demonstrated this by scoring Archer Education’s corpus, which received a 58 recognition score. While a 77% coverage score was initially recorded, it was a “lie” because 14 out of 18 lexicon terms were not clean noun phrases. This means that while the words were present, they were not grounded in a way that allowed the machine to recognize them as distinct entities.
Identifying “Blue Ocean” Content Gaps
Entity-based SEO allows practitioners to find “blue ocean” content—clusters where no one currently owns a clear answer in the knowledge graph. By comparing your entity coverage against competitors, you can identify topics they have completely ignored. This identifies high-value, low-competition opportunities where your brand can become the primary authority.
The comparison process involves crawling competitor sites, pulling their schema, and comparing their entity scores against your own. For example, if a competitor has a higher recognition score in a specific cluster, you can find the specific sub-entities they failed to define. Mapping these gaps turns technical data into a practical, actionable SEO content strategy.
A custom lexicon transforms an entity audit into a production schedule
Turning an entity audit into a production schedule requires the creation of a custom lexicon. This lexicon defines industry-specific terms, such as “stealth applicant,” ensuring they are properly mapped to recognized entities. Once defined, these terms can be integrated into your JSON-LD and content strategy to fill the identified gaps.
Python scripts can automate this process by calling the Natural Language API and serializing the results into a CSV entity map. This map serves as a visual inventory of your brand’s content gaps. Once the gaps are mapped, the results should be published to an editorial calendar to sequence content production based on the missing entities.
An Example of Using Entity Mapping for Higher Education
Suppose Sarah, an SEO manager for a university, wants to improve her brand’s visibility in AI search. She extracts her site’s JSON-LD and runs it through the Google Natural Language API, receiving a recognition score of 58. She discovers that the term “stealth application” is marked as “unrecognized” because Google only sees the word “applicant.” To fix this, Sarah adds “stealth application” to her custom lexicon, defines it clearly with surrounding context, and creates three new pages explaining the concept.
If Sarah’s initial recognition score had been 80 or higher, she would have skipped the new content creation and instead focused on refining the “partial” recognition of her existing terms. Because her score was low, the focus shifted to building out entirely new “unrecognized” clusters to compete with others.
Google has not announced pricing for high-volume corpus processing via Natural Language API
- Google has not announced the specific cost or pricing for high-volume corpus processing via the Natural Language API.
- The exact number of new Machine IDs added to the Knowledge Graph monthly has not been disclosed by Google.
The biggest risk in modern SEO isn’t that Google won’t find your page, but that it will find it and fundamentally misunderstand what you do. By using semantic triples and N-quads, SEOs provide the “relational glue” that allows an LLM to connect a brand to a specific niche rather than a generic category. If your entities aren’t recognized by the Knowledge Graph, your coverage is functionally zero for AI retrieval.
Frequently Asked Questions
What is the difference between “strings” and “things” in SEO?
Strings are the literal words used in a text, while things are the entities that the Knowledge Graph understands as distinct concepts. Moving from strings to “things” helps Google understand the underlying meaning of your content rather than just the character sequences.
>How can I measure a baseline of entity discoverability?
To measure a baseline of entity discoverability, you should extract the existing JSON-LD and HTML from your site. Run that corpus through the Google Natural Language API to see which entities are recognized, partial, or missing.
What is a semantic triple?
A semantic triple consists of a subject, a predicate, and an object. For example, Archer Education (subject) offers (predicate) enrollment marketing (object) creates a linked relationship between those three entities.
How do I identify “blue ocean” content opportunities?
To identify “blue ocean” content opportunities, you must identify entity clusters that currently have no one owning a clear answer in the knowledge graph. This is done by comparing your own entity map against a competitor’s crawl results.
How do I use Google’s Natural Language API for SEO?
To use Google’s Natural Language API for SEO, you can send a corpus of text or schema to the API to get back a match status and entity recognition scores. This reveals exactly what Google’s machine learning models “see” in your content.
What is the risk of using vague language in my content?
The risk of using vague language is that Google may fail to recognize the specific entity, such as seeing “applicant” instead of “stealth applicant.” This causes you to lose context in AI search results because the machine lacks the necessary surrounding information.
Based on Search Engine Land’s How Entity-Based SEO Can Reveal Hidden Content Gaps.