Entities and topical authority

Borislav Donchev

Author

Borislav Donchev

CEO & Founder of Back2Rank
Recovers websites after Google penalties, hacker attacks and failed migrations. 150+ recovered sites and 70+ removed penalties over more than 7 years. He has worked on cases where the cause of a traffic loss was an external attack, not an internal mistake.

Check out more posts

An entity is a thing rather than a word: a person, a company, a place, a product, recognized by a search engine as a distinct item with attributes and connections to other things. Google stores what it knows about entities in the Knowledge Graph. Topical authority is the industry’s term for a site being treated as a knowledgeable source on a subject, and it needs a caveat that most explanations leave out.

In plain English

A name written on a scrap of paper, versus a contact card in your phone.

“John Smith” on paper is a string of characters. There are thousands of them and the paper tells you nothing about which one this is.

The contact card is an entity. It has a phone number, an employer, a photograph, and a line under the name saying what he does. When someone mentions John from the tax office, you know which card they mean, even though those words appear nowhere on it.

Search engines moved from the paper to the card roughly a decade ago, and most SEO advice is still written as though they hadn’t.

Why this matters for your site

Two consequences, one immediate and one strategic.

The immediate one is that matching words stopped being how this works. Your business can be described five different ways across the web and still be understood as one thing, provided the descriptions are consistent enough to resolve to the same entity. Or it can be described inconsistently and never quite resolve into anything.

The strategic one is about coverage. A site that covers one subject thoroughly is legible as being about that subject. A site covering nine unrelated things is legible as being about nothing in particular, which is the real reason scattered publishing underperforms.

How it actually works

Strings to things. Apple the company, apple the fruit, and Apple the record label share a string and are three entities. Working out which one a page or a query means is entity recognition, and search engines do it from context rather than from the word.

The Knowledge Graph. Google’s store of entities and the relationships between them. It’s what sits behind knowledge panels, the information boxes that appear beside results for recognized people, companies, and places.

How a business becomes recognizable. Consistent naming, a page that plainly states what the organization is, structured data describing it, and being described the same way on other sites. None of this is a trick. It’s removing ambiguity so there’s one thing to recognize rather than several near-misses.

Topical authority, and what Google actually documents. In May 2023 Google described a system it calls topic authority, built to work out which expert sources are helpful for newsy queries in specialized areas such as health, politics, or finance. It named three prominent signals: how notable a source is for a topic or location, influence and original reporting measured by how often other publishers cite work you broke first, and source reputation including a history of high-quality reporting or recommendations from expert bodies. Google gave the citation signal a visible form in 2022, the Highly Cited label, which marks stories other news organizations reference often.

The caveat. That document is about news. The industry generalized the phrase to all websites, and the generalization is reasonable but undocumented. Anyone presenting a topical authority score for a plumbing company is extrapolating, and it’s worth knowing that’s what they’re doing.

What generalizes safely. The three signals Google named aren’t news-specific in spirit: being genuinely known for something, producing original work others have reason to cite, and having a record rather than a launch. Those hold up as a description of what ranking systems are built to reward regardless of sector.

What this looks like in practice

A company blog runs on whatever the marketing team found interesting that month. A post about the office move, two about industry news, one recruitment piece, three about the product, and a Christmas message.

Nothing on the site establishes what this company is expert in. Each post is fine. Together they describe an organization with no subject.

The alternative isn’t more posts. It’s twenty pages that completely cover the one operational problem this company solves, linked to each other, containing numbers and cases nobody else has. That version gets cited by other sites, which is the observable form of the thing everyone is trying to buy.

What people usually get wrong

“Topical authority is a score we can measure.” No such score is published for websites generally, and the tools reporting one are producing their own estimate.

“Publish a hundred articles to build authority.” Volume without original work is what Google’s guidance names as scaled content abuse. The signal Google actually described was being cited, not being prolific.

“Entities are just keywords with extra steps.” Keywords, meaning the words people type into search, are strings. Entities are the things those strings point at, and several strings can point at one entity.

“We need a knowledge panel.” It’s a display, not a requirement, and there’s no application process.

What you can do about it

Name yourself the same way everywhere. One legal name, one trading name, consistent across your site, your profiles, and anywhere you’re listed. Inconsistency is the most common reason a business fails to resolve into one entity.

State plainly on your site what the organization is, where it is, and what it does, and describe it in structured data as well.

Choose a small number of subjects you can genuinely claim and cover them completely, rather than covering many subjects partially.

Produce work others have a reason to cite. Original data, a real case, something first. That’s the signal Google named, in its own words, in the one document it published on this.

Link related pages to each other so the coverage reads as a body of work rather than as scattered posts. The quality of what sits inside that structure is content optimization, the work of making each page answer its question properly.

Words you will hear

Entity. A distinct thing a search engine recognizes, rather than a word it matches.

Knowledge Graph. Google’s store of entities and the relationships between them.

Knowledge panel. The information box shown in results for a recognized entity.

Topical authority. Being treated as a knowledgeable source on a subject.

Entity recognition. Working out which thing a word or page refers to.

Topic cluster. A group of related pages covering one subject between them.

For E-E-A-T, authoritativeness and YMYL, see What is E-E-A-T? For SERP, query and search console, see What is SEO?