On January 29th, Christine Schrader and I did a webinar for Conductor.
I mentioned “optimizing for entities.” This is an SEO activity I’ve been quite successful with in recent months. I gave a brief run down of this concept in the webinar, but wanted to expand on it with this blog post.
To understand the concept of entity optimization, we should start at the beginning.
Google uses different search models in their algorithms. One model, semantic search, was blended into Google’s secret sauce with the Hummingbird update. While the semantic search model wasn’t created by Google, it’s pretty interesting to see this wholehearted incorporation. It’s like discovering that adding chocolate syrup to your milk can create an even more delicious result.
The concept of semantic search is different than the contextual models Google used in the past. As SEOs, we are familiar with Google’s reliance on keywords for data retrieval. The old school search engine optimization task is simple – if you want to rank for a search query, that keyword must be in your document. But this is too simple for 2019. Through the years, Google added layers/models to improve the relevance of their results. Keywords are still part of the mix, but if 100 documents have the same keywords, how can Google tell which is the better document for their searchers? At the end of the day, keyword optimization can only help Google so much.
So Hummingbird, with its inclusion of semantic search, was Google’s next big upgrade towards understanding natural language queries. Maybe the pitch in the Google offices went something like this:
Engineer #1: “What if Google Search simply knew certain things about certain objects, so it could better serve results on what it knows?”
Engineer #2: “But keywords already help our search engine make that match!”
Engineer #1: “But if someone types in ‘Philadelphia Eagles,’ it’s now possible they’ll get a page about an Eagles sighting in Philadelphia. But obviously the searcher wanted info about the football team.”
Engineer #2: “So if we can give Google a new brain, where it knows things like ‘Philadelphia Eagles means NFL football team’, it can contribute to producing more relevant results for that searcher!”
Engineer #1: “Yes, but this is a big undertaking. This will have to be a really big brain. And it will need to always be learning, with accurate information.”
Now, to be clear, this pretend conversation is a simplistic overview on why semantic search is helpful. It helps Google better match content to the intent behind the query. There’s more to it – the semantic search model actually helps discover relationships between different queries. Try this experiment:
- Open an incognito window and search for “Brady Bunch.” Just to make sure you’re sending signals, do another search for “Greg Brady.” Finally for your third search, enter “history of brady.”
Your results are probably a mix of Brady Bunch results, and maybe a couple about Tom Brady, the history of the name, the town of Brady, and so on.
- Close that window, and open a new incognito window. Search for “NFL football.” Then search for “New England Patriots.” Finally, repeat the “history of brady” query.
This time you’ll see you’re getting even more Tom Brady results. Google is understanding your queries and how they tie together. That’s the power of this “other brain” and its understanding of entity relationships.
Let’s break it down a little more.
What Are Search Entities?
For Google, a search entity is an object that it knows about. As a user, we don’t know everything Google knows about. We don’t have a good window into their collection of entities. They have the Google Knowledge Graph Search API, but sadly I find it is not comprehensive.
But sometimes Google will show us some of the things they know right in the knowledge panel.
The information on the right indicates things Google knows about the Philadelphia Eagles. They know the head coach. They know the fight song. They know the quarterbacks. But I believe this is just scratching the surface.
As part of the semantic search model, Google can theoretically fill in the blanks. If Google knows the Philadelphia Eagles are an NFL team and they know that Doug Peterson is the coach of the Philadelphia Eagles, then they can determine that Doug Peterson is an NFL coach.
(If this stuff interests you as much as me, and you’d like to get into the weeds, do some research on Semantic Triples and Triplestores – you’ll discover a more scientific explanation of my example above.)
How Do I Optimize For Entities?
Now that you know what an entity is, we return to my concept of “optimizing for entities.”
What I have discovered (with many, many other SEOs) is the more you include relevant entities in your copy, the better your webpage performs in the rankings. If you wrote about the Philadelphia Eagles, but hardly referenced related entities, you may not get the best ranking performance. It’s as if Google trusts this content more because it triggers certain familiar entities. Please note – I am speculating on this one, as Google has not told us anything of the sort. But in my experience, there is an enormous correlation.
But you can’t have the traditional “keyword stuffing” mindset here. Don’t start stuffing in any entities hoping you’re hitting on something in Google’s brain. That will just be spammy. Instead, carefully craft your copy around the entities. Sometimes these entities can improve the topic of your website, suggesting things you hadn’t thought to write about before.
Look at competitors who are ranking above you. Scan their copy and see if you recognize possible entities that are on Google’s radar. To improve in this task, you can look at entity extraction tools. Google themselves have one – scroll down on that page and enter some competitor text in the Try the API field. This is a bit of work without a tool that scales with the API. For my needs, I’ve found Clearscope to be that tool for me.
At Greenlane, we are at the point where no piece of content written goes online without a thorough review through Clearscope (including this article).
A Crucial Consideration
If you have an article, and know what you want it to rank for, make sure you review the SERPs (search engine result pages) for that keyword. Try to identify what type of intent Google thinks the keyword suggests. For a phrase like “what is a cms,” the results show that Google thinks this is an informational query. They are showing pages that help users identify the definition or value of a CMS. They are not ranking homepages.
If your goal is to get your company’s homepage ranked here, entity optimization will not likely help unless your homepage is in line with the rest of the results. It’s not that powerful. Alternatively, since you can’t beat Google, join them; create your own “what is a cms” page, make it better than the other ranking pages, and optimize for entities. You may have a great shot to compete for this traffic.