Co-Occurrence Signals
Session 1.2 · ~5 min read
When your entity name appears on the same page, in the same paragraph, or in the same document as another entity or concept, that creates a co-occurrence signal. It is one of the most basic building blocks of entity recognition. Google literally counts how often "Entity A" appears near "Topic B" to determine relevance. This is not theory. It is how the Knowledge Graph was built from web crawl data.
How Co-Occurrence Works at Scale
Google crawls billions of pages. For every page it processes, its NLP models identify named entities and concepts. When two entities or an entity and a concept appear on the same page, that registers as a co-occurrence event. When this happens across multiple pages from multiple domains, a statistical pattern forms.
The pattern becomes an association. The more pages where "Jane Smith" and "entity SEO" co-occur, the stronger the system's confidence that Jane Smith is associated with entity SEO.
Distance Matters
Not all co-occurrences carry equal weight. The proximity of your entity name to the topic term affects signal strength. Appearing in the same sentence is stronger than appearing in the same paragraph, which is stronger than appearing on the same page.
| Proximity level | Example | Signal strength |
|---|---|---|
| Same sentence | "Jane Smith specializes in entity SEO." | Strong |
| Same paragraph | Paragraph discusses entity SEO; Jane Smith is mentioned three sentences later. | Medium-strong |
| Same section | Article section about entity SEO; Jane Smith appears under a different heading. | Medium |
| Same page | Jane Smith is in the author bio; entity SEO is in the article body. | Weak-medium |
| Same domain | Jane Smith has a profile page on a site that also covers entity SEO. | Weak |
This is why author bios at the bottom of articles provide weaker co-occurrence than having your name woven naturally into the content itself. Both matter, but proximity amplifies the signal.
Volume and Diversity
A single co-occurrence is a data point. The system needs multiple data points from multiple sources before it forms a confident association. Two factors determine when co-occurrence becomes meaningful:
Co-occurrence builds associations through repetition across diverse sources. Ten mentions on one blog are worth less than one mention on each of ten different domains.
Volume: How many total co-occurrence instances exist. More is better, but with diminishing returns from the same source. The first 5 co-occurrences on your own website establish a baseline. The next 50 on the same site add less marginal value.
Diversity: How many different domains and platforms carry the co-occurrence. This is more important than raw volume. If your name and your topic co-occur on 30 different domains, that is a much stronger signal than 100 occurrences on a single domain. Diversity indicates that the association is not just self-declared.
Natural vs. Forced Co-Occurrence
Google's NLP models can detect unnatural keyword insertion. Stuffing your entity name next to your target topic in awkward, repetitive patterns is counterproductive. The content must read naturally. A human reader should not notice that you are deliberately creating co-occurrence signals.
Good co-occurrence reads like this: "In her recent talk at SearchLove, Jane Smith outlined a framework for entity SEO that prioritizes structured data and knowledge graph optimization over traditional link building." The name and topics co-occur naturally in a descriptive sentence.
Bad co-occurrence reads like this: "Jane Smith entity SEO expert discusses entity SEO strategies for entity optimization." That reads like spam to both humans and algorithms.
Where to Build Co-Occurrence
You can create co-occurrence signals on properties you control and on external properties. Both matter, but the mix should favor external sources over time.
- Your website: About page, homepage, blog posts, service pages. Your name should appear near your core topics on every key page.
- Social profiles: LinkedIn headline, Twitter bio, YouTube channel description. These are crawled and indexed.
- Guest content: Bylined articles on other sites create co-occurrence on domains you do not control.
- Interviews and podcasts: Show notes, transcripts, and episode descriptions create co-occurrence on the host's domain.
- Directory listings: Industry directories with your name + category + description.
The next session covers co-citation, which is the external validation version of co-occurrence. Co-occurrence is the signal you build. Co-citation is the signal others build for you. You need both.
Further Reading
- Co-Occurrence and Co-Citation in SEO (Search Engine Journal)
- Co-Occurrence and Co-Citation: Prediction for SEO (Moz)
- Entity-Based Search: How Context Creates Meaning (Search Engine Land)
- Entity SEO: Signal Types (Kalicube)
Assignment
- Pick your top 3 target topic associations from your Recognition Blueprint.
- For each, search [your name + topic] in Google. Count how many unique pages create a co-occurrence between your entity name and that topic.
- Repeat the search for a competitor. Compare the co-occurrence density.
- Identify the top 5 pages on your own website where co-occurrence should exist but currently does not. Plan edits to add natural co-occurrence to each.