Knowledge Graphs - Brain Behind the shift from SEO to Discoverability

What is a Knowledge Graph?

A knowledge graph is a structured semantic network of information. It stores data, its meaning and relationships. This knowledge graph is built on triples. A triple consists of three parts - Subject, Predicate and the Object. 

To understand, lets consider an example - the sentence is ‘Michelangelo carved The David’

So here, the Subject (Node) is the entity - Michelangelo

Predicate (Edge) is the relationship - carved

Object (Node) is the related entity - The David

Nodes are the things and Edges are the connections.

By connecting billions of such triples, search engines create a web of human knowledge. Search engine understands that if you say “the guy who designed the beautiful white statue of David”, you are talking about Michelangelo.

What is the meaning of building a your own knowledge graph?

Building you own knowledge graph has two interpretations -

Public knowledge graph - Here you don’t build a graph on your site, rather you signal your place in Google’s knowledge graph. 

Private knowledge graph - Many organisations are now building their own knowledge graph to power their AI chatbots or their recommendation engines.

For the sake of this article, lets stick to public knowledge graphs.

Strategy to build a Public Knowledge Graph - Schema First Architecture

When we are building a public knowledge graph, we do not upload a graph to google. Instead, we build a machine readable “map” on your own site using the JSON-LD Schema Markup. 

JSON-LD is a snipped of code that tells google: “This isn’t just a string of text, this the organisation name, this is the name of the founder, this is their LinkedIn page” Basically, it makes your site readable in a structured manner. (More about JSON-LD later)

Some of the steps you may take are -

Establish a Root- Create /about or /organisation page that acts as a single source of truth for your business

Establish Entities - Give everything important - your brand, your authors, your products a unique stable @id. (A unique URL) This ensures google does not confuse your company “node” to another company or product with a similar name.

Establish Relationships - Use the author property to link an article to a person AND the publisher property to link it to your organisation. Another example could be - Use offers to link to specific Service entities, and founder to link to a person entity.

Use SameAS: This is the ID card of the web. Link your website entity to your official social media and other knowledge bases like Linkedin, Twitter, Medium, Wikipedia etc.

Writing for Discoverability - Entity First strategy to be in Sync with the Knowledge Graph

In 2026, We do not write with keyword embedding in mind. We write for entities and context. 

Be Explicit: We no longer build long stories, instead come to the main point. Also, we must be precise and factual in our approach. Example, we don’t say “Our Founder started this company in the later part of 2017”, we say “Our Founder Itika Singhal started Cobaltqube Media in September of 2017”

Primary and Secondary Schema: Every article that you write should clearly define its primary entity using an “About” schema and secondary entity using “Mentions” schema.

E-E-A-T Framework - EEAT framework is a crucial one for google’s knowledge graph. Since we take google as the base gold standard it applies to all search engines and LLM based chatbots. E-E-A-T stands for Experience, Expertise, Authoritativeness and Trust. Example, if your author is not a recognised node in the graph with a history of writing on that topic, your content will not be discovered.

Role of Knowledge Graphs in Generative Engine Optimisation (GEO)

GEO is the new SEO. With LLM based answer engines like ChatGPT, Perplexity, Gemini and Search Generative Experiences like AI Overviews, people don’t click on links anymore. Zero click searches are the reality of our day and age.

Hence, if you are not in the knowledge graph you don’t exist in AI. LLM based answer engines use knowledge graphs for their generative answers to:

Synthesise answers: It uses the “subject-predicate-object” fact from the graph to write a summary

Attribute Sources: It cites the “Nodes” it trusts the most. Hence giving trust and authority signals is important.

Map Intent: If a user asks a complex question, it traverses the graph to find the most relevant, connected piece of information based on user’s intent.

Your website is just one node in the massive global web. To be discoverable, start thinking about entities and context while writing your content. Also think about the schema to structure your content so that it becomes a part of the knowledge graph.