The relationship between these two terms—keywords and entities—is important in a post-Penguin SEO world. To properly optimize your website for modern search engines, including Google, you need to consider both keywords and entities and how to use them strategically in your content.
Let’s start off with definitions.
Define a keyword in 5 words or less
If we had to condense the definition of a keyword into just 5 words, it would be:
“Words people type into Google”
When it comes down to SEO, keywords are just the words we type into the search bar of any search engine. “Bars in Seattle” is an example of a keyword (or keywords). “Divorce Lawyer” is one expensive keyword if you’re running a Pay Per Click campaign.
But what people are searching for aren’t keywords. They’re searching for answers, how to get somewhere, what product to buy, who said what famous quote, etc.
So why do SEO professionals care so much about keywords? Why do we try so hard to optimize for specific keywords when that’s not what people are searching for at all?
A big part of what people are searching for is what’s referred to as an “entity.”
Define an entity in 5 words or less
Here’s our 5-word definition for an entity:
“Specific person, place, or concept”
An entity is a narrowly-defined person, thing or idea. For example, you are an entity; I’m an entity. The city you live in is an entity, as is the model of car you drive.
Concepts can be entities, too. The entity “philosophy” isn’t a physical object or place, but it’s an entity nonetheless. Topics are entities. Your company’s brand is an entity (so is its CEO).
But people aren’t necessarily searching for entities, either. For example, they might be searching to learn something about an entity, although the answer to their question isn’t that entity itself.
How are entities and keywords different, and how are they similar?
A keyword can be an entity, and an entity can be a keyword. In fact, there can (and probably will) be multiple keywords pointing to a single entity, and vice versa. Remember: keywords are just the words and phrases people type into Google. With multiple synonyms and variants, it’s quite common for there to be multiple keyword phrases that are essentially searching for the same thing. Some people might search “attorney in Cheboygan,” while others might search for “Cheboygan lawyer” and so forth. Meanwhile, there will be multiple entities (the actual lawyers who practice in Cheboygan) that relate to those keywords.
It’s all about search intent
Search engines use both keywords and entities and the relationship between the two to try to understand search intent. If a webpage contains the keywords that a user has searched for, then that’s a good signal for Google that it might be a good result. However, because Google doesn’t want SEOs to simply stuff keywords into their content (which doesn’t supply a good user experience), they look at many more factors, including entities (among perhaps hundreds more ranking signals).
The thing about entities is that they connect in predictable ways. For example, a movie, as an entity, is connected to the various actors (also entities) who starred in the movie. Each of those actors is connected to other movies they’ve been in (even more entities) as well as the cities they live in, their pet charities, and on and on in an infinitely expanding web of connections.
Some entities are closely related—so much so that it would seem out of place if one were mentioned without the other. Therefore, Google uses entity relationships to narrow down the topic of a webpage, and whether it provides the best answer for a search query.
How does a search engine sort all this out?
There are over 3.5 billion Google searches per day. (Click that link to have your mind blown.)
That’s over 40,000 per second. These are ridiculously high numbers that are difficult to comprehend.
All this data is constantly feeding into Google’s algorithm, teaching its computers more and more about search intent. This is referred to commonly as “machine learning.” All the time, Google’s machines are learning more about what people are searching for when they type in a certain set of keywords, and how entities are interconnected. Knowing exactly how they do this isn’t necessarily important; you can still use machine learning to your advantage for SEO.
2 ways to use machine learning to your advantage for SEO
The first strategy may seem counterintuitive (or overly-obvious to some), but it can be surprisingly powerful:
To rank better on Google, write for people, not machines
How could purposefully avoiding thinking about keywords and entities provide better search results? Because when you focus on providing the best experience for actual people, then the entities and keywords will come naturally and your SEO marketing will improve as a result. The synonyms, variants, and relevant entities will be a natural part of your content, and Google will pick up on that. After all, Google is most interested in understanding how people communicate with other people, not how people communicate with computers. And every passing second the search engine gets better at differentiating content that is written for computers and not people.
Disambiguate by using multiple keyword synonyms and variants (and not keyword-stuffing) and by researching secondary keywords and entities
There’s a saying in marketing: “The more you tell, the more you sell.” This works for SEO as well, because by diving deeper into the content, you give Google a better chance of understanding the topic at hand. You “disambiguate” the topic by providing multiple closely related entities and variations of keywords. Google expects certain entities and keywords to go together, and some keywords can have multiple meanings. Including the secondary keywords and related entities will help narrow it down, making your content all the more powerful. Essentially, you want to make sure all your bases are covered, and your content has adequate depth to it.
Keyword stuffing (when you use the exact same keyword over and over again throughout the content), on the other hand, does nothing to help disambiguate the topic for Google. It ends up being redundant and a waste of space, only getting in the way of a positive user experience (and could even get you in trouble with a Google anti-spam penalty).
When you combine these two strategies, you’ve just created the double-whammy: content that is helpful and relevant to what people are searching for, and that has a really good chance of ranking for the keywords people type into Google.
Content that provides the best answer is better than content that comes from a website with high PageRank
Keyword research is still a powerful tool for SEO (it helps you understand what people are searching for, the language they use, and how their search behavior is changing and evolving). It’s even more powerful when you combine it with in-depth research into topics and entities. Ultimately, the best way to rank well on Google is by providing content that provides the best answers and user experiences for actual people.