Entities and SEO - What is it and why should you bother?

Understanding entities is one of the most crucial things in SEO, if not the most, in 2020. Unless and until you’re a newbie or you’ve been living under a rock, you probably know bits and pieces about entities in search. But to be honest, I won’t blame you if you do not know/have not read much about entities because it didn’t get its share of limelight for a long time.

After all, people still love to discuss elaborately the infamous Penguin update and whether content length affects rankings, but not this.

Entities in SEO or “Entity SEO” has been a much lesser discussed topic until mid-2019. I can not stress enough on the importance it holds about how search works today and will work in the near future.

And how on earth do I know this? It’s because of these 2 Google patents - Ranking search results based on Entity metrics & Related entities. In this article, we are going to discuss a few points from this patent with inputs from Nitin, Lily, Kevin, and Andrea.

What are entities?

Entities are distinct things or objects that have an independent existence and share a relationship with other entities in a meaningful way. An entity can be anything from a non-tangible idea to places, people, events, etc.

Google defines entities as “A thing or concept that is singular, unique, well-defined, and distinguishable.” 

Entities are language agnostic, that is, it is a universal concept without any bounds on language or spelling. They use the concept of nodes and edges (PageRank much?) to demonstrate entities and their relations with other entities, where entities are the nodes and the relationships are the edges.

If you’re still confused, let’s take a real-world example of a Ferrari car. A car is an entity which comprises many other entities like the chassis, windows, tires, etc. It is also a part of a bigger entity - automobiles. And since an idea is also an entity, the idea of a “car running on water” can also be an entity.

Now, the second layer to this is the relation between two or more entities. The car, in the discussion here, is related to Ferrari (another entity) which is the brand of the car. Similarly, all the other entities mentioned above are related to the car entity in a very meaningful way. Note that all the entities are distinct and can exist independently.

Let us shift back to the world of SEO once again. Your website is an entity composed of other entities such as articles, product pages, logos, images, etc. An article, in turn, has several entities linked to it which are distinct words and phrases that it is composed of. And this will go on and on until what you’re left with a mesh of inter-related entities.

Did Google invent the concept of entities?

Definitely not. Entity recognition or entity extraction is a concept of machine learning that Google has simply implemented in its algorithm. In fact, Natural Language Processing concepts like named entity or information extraction can be dated back to as early as the 1990s. 

Named Entity Recognition (NER) is a machine learning technique that is used to extract information from text. This extracted information is then identified and classified into named entities. These entities can either be pre-defined and very generic or can be new and specific, in which case the NLP model can be trained to classify them.

We come across loads of applications today in our daily lives that use NER in such a subtle way that we don’t even realize it. Have you noticed how Gmail asks you whether you forgot to attach a file if you try to send an email (which has the term “Please find attached” written in it) without attaching a file? Yes, that’s probably NER working in the background. Google’s search algorithm most certainly uses this NLP model to understand entities and their relations.

Search results based on entity metrics 

Google, in its patent, states that a score is determined by combining certain entity metrics to determine what to rank. Also in certain cases, it uses a “weighted combination of entity metrics, where weights are assigned to each of the metrics based on the category or entity type of search results.”

What are these entity metrics?

Google mentions 4 such metrics that it uses to determine a score:

• Relatedness metric
• Notability entity type metric
• Contribution metric
• Prize metric

The relatedness metric

The relatedness metric is determined based on the co-occurrence of 2 or more entities. What this means is if 2 entities appear multiple times on the web together (on trusted sources), Google can identify a relationship between them. 

For example, Donald Trump [entity] is the President of this United States [relationship].

Thus, when you search for something like “who is the President of the United States”, Donald Trump appears as shown in the image.

This is a sort of explicit way of using entities. Google also uses entities implicitly where they try to validate featured snippets of the same query in reverse.

Thus, when you search for “who is Donald Trump”, a featured snippet appears from its knowledge graph which states that he is The President of the United States.

This is what Andrea Volpini, the CEO & co-founder of WordLift, has to say about the importance of entities in search:

"As Google uses Entities to match user intent with search results, we can inform our content strategy around Entities. We call it the Entity-based content model; basically a way to write content that allows both humans and machines to share meanings. At a very practical level, Google uses entities in an explicit way with a knowledge panel and answers like "who is the Ceo of WordLift?" But even more importantly Google uses entities implicitly to validate statements behind featured snippets like "what does Andrea Volpini do for a living?"

So whenever you’re writing a piece of content, it is not only essential that you include all possible entities related to your topic but also you establish the relationship between them in a meaningful way.

Notability metric

The notability metric basically talks about how “notable” an entity is among other entity types in a particular domain. I’ll quote Google’s example here.

Consider the domain “books”, where “short stories”, novels, “non-fiction” are all entity types. These entity types may be ranked in the order of Novels > non-fiction > Short stories as novels are probably the most notable and short stories, the least. So, for a particular domain, the collection of entity types shall be ranked by their notability, popularity, frequency of occurrence, etc.

Contribution metric

The contribution metric of an entity is determined by external factors such as reviews, ratings, fame publications, a reputed online coverage, and the likes. So, basically a backlink or a mention from Hubspot(entity) will be more effective than some random XYZ site because its entity contribution is higher in that domain.

Prize metric

As the name suggests, the prize metric is determined by awards and recognition and the entity has achieved. For example, the Oscar-winning movie “Parasite” will be a more weighted entity compared to any other movie in 2020. The more prizes linked to an entity the higher it is ranked among several entity types.

Why this new approach?

If you notice all of google’s updates lately including BERT serves the sole purpose of understanding text better. This fuels Google’s leap into being an “Anwer engine” (as some like to call it) which is more powerful than being just a search engine.

Entities only strengthen the foundation of this entire process by re-indexing the entire web and forming a sort of entity graph. This helps Google to accurately nail SERP features like featured snippets, the knowledge graph, etc.

Voice search is another area where the role of entities can’t be ignored. Most of voice search answers are fulfilled by featured snippets, which in turn relies on entities to validate the answers.

An entity-first approach also makes it easier to shift away from the link graph structure. We all know how important external references (backlinks) play when it comes to ranking your site. With understanding entities and their relations, identifying (and thus ranking) quality & relevance of content will be much easier and it no longer needs to rely on just links.

This approach also solves the issue of language. Since entities are not limited by language, it’ll be much easier to understand what text means in any language.

Here’s what Kevin Indig, VP of SEO & Content at G2, has to say about how important it is for us to understand entities:

“Entities already have been impacting SEO since Google added the knowledge graph in 2012. They serve as the basis for many SERP Features, even more so since the Topic Layer came on top of the Knowledge Graph.

Google's understanding of entities and their relationships seems really good from my point of view, which means it can assess content relevance for a query much better. For us, it means that we need to factor in entities when optimizing and writing content.”

Optimize your content in the world of entity-based indexing

Let’s address the elephant in the room, shall we?

All these boil down to one thing in my opinion - writing genuine, helpful, and relevant content that matches the user intent (more on schema and E.A.T. later).

Here are a few pointers that will help you optimize your content better:

The statement “If the content is king, then context is queen” is now truer than ever. When you finalize on the topics or keywords you want to rank for, start by understanding what a user might actually be looking for when they search. Write deeper, more meaningful content so that the user coming on to your platform doesn’t have to go anywhere else after reading your content.

How do you find what “entities” or subtopics to include in your content? Well, there are a few ways:

Wikipedia

Look at what Wikipedia has to say about your chosen keyphrase. For all we know, it is one of the strongest sources of Google as far as their entity graph and knowledge panel are concerned. See if you can add more “relevant” subtopics in your content or not.

PAA

Questions in the People Also Ask section is a clear indicator of what entities Google relate to with your chosen keyphrase. Include them in your content and try providing a comprehensive answer to each.

Entity Explorer

Entity Explorer is something that I personally use and love. It is an amazing tool that shows you related entities for any key phrase that you search for. More on this later in the article.

Google's NLP API

Use Google’s NLP API to see how Google views and identifies entities in your articles. This is what Kevin said about content optimization with entities:

“Google's NLP API is a nice start to become aware of what entities a piece of content covers and how high their salience is. Entityexplorer.com visualizes that very nicely and can serve as part of a content briefing. Rebuilding a whole site structure after an entity taxonomy works really well, too.”

WordLift

WordLift is another cool WP-plugin that helps you optimize your content with entities and form an intricate entity graph. How does WordLift optimize your content with entities? This is what Andrea said:

“WordLift is designed to publish linked knowledge graphs that help search engines and smart digital assistants understand the content we care about. Having your own personal knowledge graph helps you have control over your data and informs platforms like Google, Bing, and Amazon. More importantly, you can use your own graph for improving the content recommendations (helping readers discover content on your website), or the search on your own site. AI is all about data and knowledge graphs are meant to transfer human knowledge (in the form of structured content) to machines of all kinds.“

Check out this SERP analysis by Andrea to have a look at how machine learning aids to entities.

It is important to mention here that the internal links on your site will play a huge part in how Google sees you as an entity, in my opinion. Also if you’re building links from external sites, make sure the links are topically relevant. Remember, quantity > quality (contribution metric).

Entities help Google to understand texts as humans would. So while writing content, just try to be as natural as possible. If you write a genuinely comprehensive and helpful piece of content, a countless number of relevant entities will get included in your content. This is what Nitin Manchanda, Global Head of SEO at Omio, approaches entities:

“Not much difference for us here at Omio. I’m a big fan of building products for users and therefore, the focus was always on understanding different personas and the intent. Once we know what our users are looking for, we (the SEO team) sit together with design, product, and engineering to brainstorm on ideas to map these needs with great content elements on our landing pages. That’s how we connect multiple entities at Omio, in a nutshell.”

I guess now you probably have a better idea of what entities are, why do they event matter, and what you can do about it. I'll pen down another article where I'll discuss Google's "Related entities" patent in more detail and a walkthrough of how to approach writing better content.

Entities, E.A.T. and Schema - The holy trinity

In order to completely understand and implement entities, you have to factor in E.A.T. and structured data. If you're not using structured data to mark up your webpages and content, you're losing out on a lot. Schema helps search engines to understand your content and niche better by removing ambiguity (if any). I'll discuss this in much more detail in my next article.

Lily Ray, SEO Director at Path Interactive, said this about how entities are affecting search and how you can control how Google sees you as an entity:

"Being a recognized entity within Google's Knowledge Graph can help your SEO performance in a number of ways. For one, it enhances the way your brand name and/or personal name appears within the search results, which you can then further optimize and enhance by claiming your Knowledge Panel. Beyond that, if Google has a confidence about who you are as an entity, this can contribute to your E-A-T (expertise, authoritativeness, and trustworthiness) and your ability to rank for more competitive queries in your niche.

You can control the associations of your entity by ensuring you have leveraged structured data as much as possible to indicate who you are, what credentials you have, and the other entities with which you associate. You can solidify this by using the sameAs property in Schema.org to link to your Knowledge Graph URL. You can also ensure that you have a strong backlink profile with trusted citations pointing back to your brand."

Check out this article by Lily on the Search Engine Journal on how to use structured data to support E.A.T. for more information.