From Keywords to Entities: How Search Changed
Google stopped being a keyword-matching engine years ago. The Knowledge Graph, which launched in 2012 and now contains billions of entities, fundamentally changed how search works. Instead of just matching words in your content to words in a query, Google tries to understand what things are and how they relate to each other.
An entity is a thing — a person, place, concept, organization, product — that's uniquely identifiable. "Apple" the company and "apple" the fruit are different entities with different Knowledge Graph IDs. When you search for "apple stock price," Google doesn't need the word "stock" on Apple Inc.'s entity page. It understands the relationship between the Apple entity and stock market data.
How the Knowledge Graph Connects Information
Think of the Knowledge Graph as a massive web of nodes (entities) connected by edges (relationships). Google builds this from multiple sources:
- Wikipedia and Wikidata — the backbone of entity definitions
- Structured data from websites (Schema.org markup)
- Google's own databases (Maps, Books, Shopping)
- CIA World Factbook, Freebase (now absorbed), government databases
- Patterns extracted from crawling billions of web pages
When Google encounters your content, it doesn't just index keywords. It tries to identify which entities you're discussing and how they relate to each other. This is why two pages can rank for the same query even when they use completely different words — they're both about the same entity and relationships.
Entity Salience and Its Ranking Impact
Google's Natural Language API has a concept called "entity salience" — a score from 0 to 1 indicating how central an entity is to a piece of text. You can actually test this yourself with the API.
Pages that rank well tend to have high salience scores for their target entity. That means the entity isn't just mentioned — it's the central topic of the content. A page that briefly mentions "PostgreSQL" while mostly discussing general database theory will have low salience for the PostgreSQL entity. A page deeply focused on PostgreSQL configuration, performance, and specific features will score much higher.
Practical Entity Optimization
1. Establish Your Own Entity
If your brand doesn't have a Knowledge Panel, you're invisible to entity-based search. Steps to establish one:
- Create a Wikidata entry for your organization (it's easier than Wikipedia and feeds the Knowledge Graph directly)
- Claim your Google Business Profile
- Use consistent Organization schema markup across your site
- Get mentioned (with consistent naming) on authoritative third-party sites
- Build a clear "about" page with structured data that matches your Wikidata entry
2. Write Around Entities, Not Just Keywords
When creating content about a topic, identify the core entities and their relationships. Writing about "Docker containers" means you should naturally reference related entities: images, Dockerfiles, registries, container runtimes, Kubernetes, namespaces. This isn't keyword stuffing — it's providing the context Google needs to understand your content's depth.
I'd recommend running your draft through Google's Natural Language API (it's free for small volumes). Check which entities it identifies and their salience scores. If your target entity isn't the highest-salience result, your content might be too broad.
3. Use Structured Data to Define Entities
Schema.org markup is your direct communication channel with Google's entity understanding. At minimum:
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Your Company Name",
"sameAs": [
"https://www.wikidata.org/wiki/Q12345",
"https://twitter.com/yourcompany",
"https://www.linkedin.com/company/yourcompany"
],
"url": "https://yoursite.com"
}The sameAs property is especially powerful for entity disambiguation. It tells Google "this is the same entity as these other references," helping consolidate your entity signals.
Semantic Search and Topic Depth
Google's BERT and MUM models understand language at a semantic level. They don't just count words — they assess whether your content genuinely covers a topic.
For any target entity, there's a cluster of related entities and concepts that a comprehensive treatment should touch on. Missing important related entities signals shallow coverage. Tools like InLinks, Surfer SEO's NLP analysis, and MarketMuse map these entity clusters and show you what's missing.
But don't just bolt on paragraphs to hit entity checkmarks. The content needs to flow naturally. Readers (and Google) can tell when you've artificially crammed in related terms. Write for genuine understanding and the entities will appear naturally.
Knowledge Panel Optimization
If you or your brand has a Knowledge Panel, you have some control over what appears there:
- Claim it through Google's verification process
- Suggest edits for incorrect information
- Your structured data influences what Google displays
- Prominent mentions on authoritative sites affect panel content
Keep in mind that Knowledge Panels are entirely Google's creation. You can influence them, but you don't control them. Google might show information from Wikipedia, your website, social profiles, or third-party databases — whatever it considers most relevant and reliable.
Tracking Entity Optimization Results
Measuring entity SEO impact is trickier than tracking keyword rankings. Watch for:
- Appearance of Knowledge Panels for your brand name queries
- Rich result eligibility increasing (entities help trigger rich snippets)
- Ranking improvements for semantic variations of your target queries
- Google Search Console showing impressions for queries you didn't specifically target but are semantically related
Entity optimization is a long game. You won't see overnight results. But over 6-12 months, sites with strong entity signals consistently outrank those relying purely on keyword-focused content. It's how Google wants the web to work, and aligning with that direction is just good strategy.