hibranwar
Course → Module 3: Entity Linking: Connecting Your Digital Presence
Session 1 of 6

The Entity Graph

Session 3.1 · ~5 min read

In Module 2, you learned to describe your entity using structured data. But structured data on your own website is only one input. Google builds its understanding of your entity by collecting signals from across the entire web: your website, your social profiles, business directories, Wikipedia, Wikidata, news articles, government registries, and more.

These signals form a graph. Not a chart or a diagram, but a mathematical graph: a network of nodes (entities) connected by edges (relationships). Your entity is one node. Every website, profile, directory listing, and mention that references your entity creates an edge. The density, consistency, and authority of these edges determine how confidently Google identifies and presents your entity.

What Is an Entity Graph

Google's Knowledge Graph is, at its core, a massive entity graph. It contains billions of entities and trillions of relationships between them. When you search for a company and see a Knowledge Panel, you are seeing Google's rendering of that entity's node in the graph, along with its most important edges.

Your entity graph is the subset of the Knowledge Graph that relates to you. It includes:

graph TD W["Your Website"] <-->|sameAs / url| LI["LinkedIn"] W <-->|sameAs / url| FB["Facebook"] W <-->|sameAs / url| TW["Twitter/X"] W <-->|sameAs / url| IG["Instagram"] W <-->|GBP link| GBP["Google Business Profile"] W <-->|citation| YP["Yellow Pages"] W <-->|sameAs| WD["Wikidata"] WD <-->|described by| WP["Wikipedia"] GBP -->|feeds| KG["Google Knowledge Graph"] WD -->|feeds| KG W -->|structured data| KG LI -->|entity signal| KG YP -->|citation signal| KG style W fill:#222221,stroke:#c8a882,color:#ede9e3 style LI fill:#222221,stroke:#6b8f71,color:#ede9e3 style FB fill:#222221,stroke:#6b8f71,color:#ede9e3 style TW fill:#222221,stroke:#6b8f71,color:#ede9e3 style IG fill:#222221,stroke:#6b8f71,color:#ede9e3 style GBP fill:#222221,stroke:#c47a5a,color:#ede9e3 style YP fill:#222221,stroke:#8a8478,color:#ede9e3 style WD fill:#222221,stroke:#c47a5a,color:#ede9e3 style WP fill:#222221,stroke:#c47a5a,color:#ede9e3 style KG fill:#222221,stroke:#c8a882,color:#ede9e3

Link Types in the Entity Graph

Not all edges in the entity graph carry equal weight. The type of link, its direction, and the authority of the source all matter.

Link Type Direction Example Signal Strength
Official website link Website → Profile Footer link to LinkedIn Strong (you control it)
Profile backlink Profile → Website LinkedIn "Website" field Strong (confirms ownership)
sameAs declaration Website → Profile JSON-LD sameAs array Very strong (machine-readable)
Citation (NAP) Directory → Entity Yellow Pages listing Moderate (third-party confirmation)
Editorial mention News site → Entity Press article with link Strong (editorial authority)
Wikidata statement Wikidata → Entity Official website (P856) Very strong (structured, authoritative)
Cross-profile link Profile → Profile Instagram bio linking to LinkedIn Moderate (reinforcement)
User-generated mention Forum/review → Entity Reddit mention with link Weak (uncontrolled)

Bidirectional Links

The most powerful pattern in entity linking is bidirectionality. A link from your website to LinkedIn is a claim: "This is our LinkedIn." A link from LinkedIn back to your website is a confirmation: "Yes, this website belongs to this LinkedIn profile." Together, they form a verified connection.

Unidirectional links are claims. Bidirectional links are proof. Google treats them very differently.

Key concept: Every link in your entity graph has a direction. A claim (outbound link) says "I am connected to this." A confirmation (inbound link from that same destination) says "Yes, that is true." Entity authority is built by converting claims into confirmed, bidirectional connections.

Graph Density and Confidence

Google's confidence in your entity is a function of graph density: how many confirmed connections exist, how authoritative the connected nodes are, and how consistent the information is across all of them.

Consider two businesses with the same name. Business A has a website, one social profile, and no directory listings. Business B has a website with structured data, five social profiles all linking back, a Google Business Profile, Wikidata entry, three directory citations, and two press mentions. Business B has a denser graph. Google can identify and disambiguate Business B with much higher confidence.

How Google Processes the Entity Graph

Google does not simply count links. It evaluates:

  1. Consistency: Does every node in the graph present the same name, address, phone, and description?
  2. Authority: Are the connecting nodes authoritative? A Wikidata entry carries more weight than a random directory.
  3. Recency: Are the profiles active and recently updated?
  4. Reciprocity: Are the links bidirectional?
  5. Relevance: Do the connected nodes make sense together? A restaurant linked to medical directories is suspicious.

The remaining sessions in this module focus on building each type of connection: website-to-profile, profile-to-website, cross-profile, and the formal sameAs declaration. Then we cover how to audit the whole graph to catch breaks and inconsistencies.

Your Entity Graph Map

Before you build links, you need to know what exists. Map your current entity graph by listing every place your entity appears online. For each, note:

This map becomes the foundation for the linking work in Sessions 3.2 through 3.6.

Further Reading

Assignment

Create your entity graph map.

  1. List every online presence for your entity: website, social profiles, directory listings, GBP, Wikidata, press mentions.
  2. For each, record: Platform, URL, Links to website (Y/N), Website links to it (Y/N), NAP match (Y/N).
  3. Count: how many bidirectional links do you have? How many are unidirectional? How many have broken or missing return links?
  4. Identify the three highest-priority gaps (missing links or broken connections).
  5. Draw the graph. Put your website at the center and draw lines to each presence. Use solid lines for bidirectional and dashed lines for unidirectional.