Start with things and relationships, not a keyword spreadsheet
A keyword list records language people use. An entity model records what the business is talking about. The organization is an entity. Each meaningful service is an entity. Founders, specialists, locations, products, certifications, and documented projects may also be entities. Questions and topics describe how people seek information about them.
Entity first does not mean ignoring search demand. It means using demand to improve a coherent model instead of producing disconnected pages for every phrase. The business should be able to draw the relationship between the company, a service, the person responsible, the location served, the evidence available, and the questions a buyer asks.
Build an approved entity inventory
List the entities the business can support publicly. Record the preferred name, type, concise description, canonical page, important attributes, related entities, public sources, owner, and review date. Mark facts that are provisional, private, outdated, or unavailable.
| Entity type | Core facts | Potential evidence |
|---|---|---|
| Organization | Public name, URL, contact, founder, operating area | About page, legal record, public profiles |
| Service | Scope, fit, process, limits, next step | Service documentation, project examples, source material |
| Person | Name, role, relevant experience, authored work | Profile page, article bylines, approved credentials |
| Location | Address or service area, availability, local context | Location page, verified business profile |
| Project | Client or category, constraint, work, date, observable result | Case study, live page, deployment or project artifact |
Do not add an entity merely because a schema property exists. Add it because the business needs to explain a real thing consistently.
Give important entities a canonical home
Choose the page that establishes each important entity. The about page can establish the organization and founder. A service page can own the full explanation of one service. A location page can own a real geographic operation. A project page can document one engagement or build.
Other pages can mention the entity and link back to its canonical home with descriptive text. This creates a comprehensible site graph. It also reduces duplication because supporting articles can answer narrower questions without repeating the complete service description.
Attach real questions to the right entity and page
Collect questions from sales, support, search data, reviews, forums, and subject matter experts. Classify them by the entity and decision they concern. A question about service fit belongs near the service. A question about a specialist's experience may belong on the person page. A complex regulatory explanation may need an article linked from the relevant service.
Prioritize questions by business importance, user need, evidence availability, and content gap. Search volume can inform the decision but should not force a page when the business lacks authority or a useful answer. One complete response is usually more valuable than several thin variations.
Use internal links to express relationships
Navigation shows the broad structure, while contextual links show meaning. A service page can link to the responsible team member, a related project, a detailed process guide, and the contact path. An article can link to the service it supports and to primary sources. A project can link back to the capabilities demonstrated.
Use descriptive anchor text rather than repeated phrases such as learn more. Keep links useful to a reader. Avoid creating a dense mesh where every mention links everywhere. The relationship should answer why the destination helps at that moment.
Maintain a shared fact and evidence layer
Store approved public facts in a source that content owners can review. The organization name, founder, contact details, service names, locations, and credentials should not drift across templates. When a fact changes, update the canonical source, visible pages, structured data, profiles, and discovery files that depend on it.
Attach evidence to claims. Record the source URL or artifact, publication date, last verification date, approved wording, and limitations. This is especially important for statistics, regulations, certifications, rankings, and time sensitive platform behavior. When evidence expires or becomes unavailable, revise the claim rather than carrying it forward automatically.
Let structured data mirror the visible graph
JSON-LD can express selected relationships explicitly. Organization markup can name the founder and logo. ProfilePage can identify the person presented on an about page. Article can connect a guide to its author and publisher. BreadcrumbList can represent page hierarchy.
Use stable absolute identifiers and URLs. Keep the markup limited to facts visible or clearly established on the page. Structured data should be generated from the same content source when practical, then validated after rendering. It is a semantic aid, not a substitute for the page.
Govern the architecture as the business changes
- Assign an owner to each core entity and canonical page.
- Record the approved name, description, facts, sources, and review date.
- Audit new pages for duplicate ownership and unsupported relationships.
- Review high risk or time sensitive facts on a defined schedule.
- Update visible content, structured data, sitemap, and internal links together.
- Retire or redirect obsolete pages with an explicit reason and destination.
A maintained entity model gives design, content, SEO, CRM, and agent workflows a shared vocabulary. That consistency is the real advantage.
Frequently asked questions
What is entity-first content architecture?
It is a way to organize a site around identifiable things and their relationships: the company, services, people, locations, projects, credentials, and customer questions. Each important entity gets one canonical home, consistent facts, and descriptive links to related entities. Articles then support that model instead of becoming an isolated pile of keywords.
How does entity-based SEO help a service business?
Entity-oriented retrieval research shows that words, entities, and their relationships can provide complementary signals. For a service business, the practical benefit is clearer ownership: the site can distinguish who provides which service, where it is available, what proof supports it, and which guide explains the buyer question. This helps people navigate and reduces contradictory pages.
How many pages should one service have?
Usually one canonical service page should own the offer, scope, process, and conversion path. Supporting articles can answer distinct educational questions and link back to that page. Create another service or location page only when it represents materially different intent, proof, or availability. Repeated near-duplicate pages weaken navigation and make factual governance harder.
What makes an internal link useful in an entity-first site?
A useful internal link connects two genuinely related page jobs with anchor text that describes the destination. A guide about crawlability should link to the technical website service and to the crawler guide, not every page on the domain. Maintain links in both directions when the relationship helps readers move from explanation to evidence or action. Audit orphan pages, repeated generic anchors, and links to superseded canonical pages.
Academic sources
These peer reviewed papers, conference proceedings, and scholarly preprints support the research and implementation guidance in this article. Each link points to the publication or an academic repository.
- Hypergraph-of-Entity: A General Model for Entity-Oriented Search (INESC TEC and University of Porto / arXiv preprint)Joint representation research for text, entities, and explicit relationships.
- Entity-Duet Neural Ranking (Association for Computational Linguistics)Evidence that entity annotations and knowledge-graph semantics can complement word-based retrieval.
- Structural Feature Engineering for Generative Engine Optimization (arXiv preprint)Preprint connecting document architecture and information chunking to generative citation behavior.
- SourceBench: Can AI Answers Reference Quality Web Sources? (UC San Diego / arXiv preprint)Source-quality framework emphasizing relevance, accuracy, accountability, authority, and clarity.