Building Topical Authority Through Entity Relationships — A Practical Guide

Topical authority is the SEO outcome where a domain is recognised by the engine — and by AI systems — as the most credible source on a topic. A site with topical authority for "personal injury law in Johannesburg" outranks individual page-level efforts on every related query in that topic cluster. The work to build it is not content volume; it is entity modelling.

This post explains what entity-led topical authority is, why it matters in 2026, the entity types that matter for a South African business, the relationship patterns that signal authority, and the practical work to build it.

What "entity" means in SEO

An entity is a thing the engine can recognise as a distinct object. People, places, organisations, products, services, concepts — all entities. The engine models each entity with attributes (name, address, founder, founding date) and relationships (works for, located in, sells, competes with).

Traditional SEO treats a page as a target for a keyword. Entity-led SEO treats the business as a node in a graph of entities, and the website as the documentation of that node. The keyword becomes a query against the graph; the page that best answers the query ranks.

The shift: pages no longer compete for keywords. Pages compete as representations of entities. The page that best represents the entity — and the entity's relationships — wins.

Why entity-led authority matters in 2026

Three reasons:

1. AI search engines rank entities, not pages. ChatGPT, Perplexity and Gemini cite entities. A business whose entity is well-modelled is cited more often than a business whose entity is poorly modelled, regardless of how many pages the second business has.
2. Google's Knowledge Graph drives many SERP features. Knowledge panels, People Also Ask, related entities, topical entity cards — all are driven by the entity model. A site that participates in the entity model gets these features; a site that doesn't, doesn't.
3. Topical authority compounds across queries. A site with topical authority for "personal injury law in Johannesburg" outranks for "what to do after a car accident in Johannesburg," "average car accident settlement South Africa," "how long do I have to file a claim in SA" — every query in the topic cluster. The lift is broader than page-by-page ranking work.

A business that invests in entity modelling gets compounding returns across the entire topic. A business that invests only in pages gets page-by-page returns.

The entity types that matter

For a South African business, six entity types matter:

1. Organization. The business entity. Name, address, contact, founders, founding date, sameAs links, social profiles.
2. Person. The founders, the leadership, the author of the blog content. Person schema with sameAs to LinkedIn, Twitter, Wikidata where applicable.
3. Place. The cities served, the suburbs served, the regions served. LocalBusiness with areaServed set to specific cities.
4. Service. The services offered. Service schema with provider (the Organization), areaServed (the Place), offers (the pricing).
5. Product. The products sold (for eCommerce). Product schema with brand, manufacturer, offers.
6. CreativeWork or Article. The blog content, the case studies, the original research. Article schema with author (the Person), publisher (the Organization).

A South African business modelling all six entity types across the website is participating fully in the entity graph. A business modelling only Organization and LocalBusiness is participating minimally.

The relationships between entities

The relationships matter as much as the entities themselves. The relationship patterns:

• Organization → Place. "Page Panther is located in Johannesburg." Schema: LocalBusiness with address.
• Organization → Person. "Penny Kruger founded Page Panther." Schema: Person with memberOf or worksFor pointing to Organization.
• Organization → Service. "Page Panther offers local SEO." Schema: Service with provider pointing to Organization.
• Service → Place. "Page Panther offers local SEO in Johannesburg." Schema: Service with areaServed pointing to Place.
• Service → Service. "Local SEO is part of the SEO service cluster." Schema: Service with isRelatedTo pointing to sibling Service.
• Article → Person. "Penny Kruger wrote this article." Schema: Article with author pointing to Person.
• Article → Organization. "Page Panther published this article." Schema: Article with publisher pointing to Organization.
• Article → Topic. "This article is about local SEO." Schema: Article with about pointing to Topic entity.
• Person → Person. "Penny Kruger worked at Company X before Page Panther." Schema: Person with knowsOf or colleague.

The relationships are encoded in the schema. Each relationship is a property on one entity pointing to another entity's @id. The graph is built by the schema; the engine reads the graph and constructs the entity model.

A practical entity-led content strategy

For a South African business, the entity-led content strategy looks like:

1. Map the entities. List the Organization, the People (founders, leaders, authors), the Places (cities, suburbs, regions), the Services (the offerings), the Products (for eCommerce), and the Topics (the subjects the site covers).
2. Model each entity in schema. Each entity gets a JSON-LD representation on the relevant page. The @id values are consistent across the site.
3. Build the relationships in the schema. Each entity's properties point to the @ids of related entities.
4. Publish content that documents the entities. Each entity should have at least one page that explicitly describes it. The Organization has an "About" page. Each Person has a bio. Each Place has a service-area page. Each Service has a service page. Each Topic has at least one article.
5. Interlink to express the relationships. Internal links are the visible expression of the entity relationships. The Organization's About page links to the People, the Places, the Services. Each Service page links to the Place it serves and the Articles about it. Each Article links to the Author and the Topic.
6. Validate the entity graph. Use Google's Rich Results Test and Schema.org Validator to confirm the schema is valid. Use a knowledge graph tool to visualise the relationships.

A site that completes all six steps over 6 to 12 months has a measurable entity presence in the engine's model. The citation work and the topical authority work compound from there.

The compounding pattern

Entity-led SEO compounds in three ways:

1. Across queries. A site with strong entity modelling ranks for every related query in the topic, not just the queries targeted by specific pages.
2. Across surfaces. The same entity model supports traditional SERPs, AI Overview citations, ChatGPT citations and Perplexity citations. One investment, multiple surfaces.
3. Across time. An entity model built in 2026 still supports the business in 2030. Page-level SEO work degrades as queries change; entity-level SEO work persists.

A business that invests in entity modelling for 12 months is investing in an asset that pays back for years. The compounding is the difference between SEO as an expense and SEO as an asset.

Common mistakes

Four mistakes to avoid:

1. Schema without relationships. Entity schema with no relationship properties. The schema is technically valid but adds no entity value.
2. Inconsistent @ids. The same entity referenced with different @ids across the site. The engine cannot link the entities.
3. Content that contradicts the schema. A page that claims the Organization is in Johannesburg while the schema says Pretoria. Trust signals collapse.
4. Building entity work without topical depth. Entity modelling is the framework; topical depth is the content. A site with perfect schema and thin content does not earn authority.

A business that invests in schema and skips the content (or vice versa) gets half the return. The two work together.

FAQ

Do I need a Wikidata entry for my business?

Only if you qualify. Most SMEs do not qualify — Wikidata requires notability. A business that has won awards, been featured in trade press, or has a notable founder may qualify. The Wikipedia notability guideline is the standard.

How long does entity-led SEO take to show results?

3 to 6 months for the schema and content foundation. 6 to 12 months for the topical authority to compound. The work is slower than page-level SEO but the returns are larger and longer-lasting.

Can I do entity-led SEO myself?

Partially. The schema work can be done with plugins (Rank Math, Schema Pro) and JSON-LD generators. The content work requires editorial capability. The strategic work — deciding which entities to model, which relationships matter — usually requires a strategist.

Is entity-led SEO only for big brands?

No. SMEs benefit disproportionately because the entity work compounds faster on a smaller site. A 20-page SME site that ships entity schema and topical depth can outrank a 200-page competitor that ships neither.

Does entity-led SEO replace traditional keyword research?

No. Keyword research informs which queries to target. Entity modelling informs how to model the business. The two work together.

If you want a second opinion on the entity work for a specific business, send us the URL and the topics you want to be authoritative on. Most entity optimisation engagements are scoped as 6 to 12 month projects, starting with the entity map and the schema foundation, then the content work layered on top.