Categories
Google Algorithm

Google RankBrain: How It Works and What You Can Do

Have you heard of Google RankBrain? 

It’s the third most important ranking signal. 

Google scientist Greg Corrado said this in a Bloomberg news story in October 2015. That was the first time RankBrain’s existence was confirmed. 

Greg Corrado, Google scientist talking about RankBrain on Bloomberg in 2015.

However, Google RankBrain isn’t really a “ranking signal” in the traditional sense. Once you understand it, you can approach your content the right way. 

Here’s what you need to know.  

What Is Google RankBrain?

Google RankBrain is a system that uses machine learning to find the most relevant pages for a search query. It has been cited as part of Google’s Hummingbird — a major algorithm change in 2013. 

Here’s how RankBrain works, in a nutshell:

1. It takes an unknown query.

2. Figures out the intent and topic behind that query.

3. It brings back the best possible results for that query.

Then, RankBrain looks at how users interact with those search results. 

Do people click? Or do they try other queries to find a result that satisfies their query? 

Data from users helps RankBrain improve and optimize search results over time. 

As you can see, it is in a constant learning process. With over 40,000 search queries happening every second, there’s a lot to learn from. 

This is especially true as voice search queries continue to increase. Voice queries tend to be different than written ones, which leaves a lot of room for interpretation. 

Also, RankBrain doesn’t only affect queries in the English language, as tweeted by Gary Illyes: 

Source: Twitter
💡Bottom line: RankBrain makes Google a better search engine by giving users what they want. 

Now that you have an idea of how RankBrain works, let’s consider the scenario below. 

RankBrain In Action

Imagine that you entered this search query into Google:

“Where should I eat?”

Me, looking for a restaurant.

RankBrain takes that query and figures out what you mean with that query. 

Do you need suggestions on the best restaurants worldwide? Within your country/state/city? Or are you simply looking for restaurants nearby? 

RankBrain obtains data from Google, analyzes it, and recommends pages based on various signals like past search queries, location, device, and content freshness. 

Let’s say that you weren’t satisfied with the results that Google provided. 

Instead of getting suggestions for nearby restaurants (which you expected), Google showed you roundup articles of the best restaurants worldwide. And some trivia quizzes too, which were irrelevant. 

So you changed your original search query into:

“Where should I eat near me?”  

Me, looking for a restaurant in a slightly clear way!

Boom. This time, you got the results you wanted. That small tweak in your query made the difference. 

RankBrain takes your behavior into account. Since you ignored the results during your initial search, it now understands that it has to do a better job at serving more relevant content in the future for your first query. 

You see, patterns in searchers’ behaviors help Google refine search results to satisfy users. If you entered the same query a few months from now, you might see different results! 

Here, we appreciate the machine learning aspect of RankBrain — the ability of an algorithm to teach itself from experience without being programmed. 

By using machine learning, Google can show users results from a query even when that query was never searched for before. Over time, results improve as RankBrain continues to learn from users. 

💡Interesting fact: In 2015, RankBrain was analyzed only on less than 15% of queries. The following year, it was already used in all queries. 

Pre-RankBrain Era

Back in the day, website owners could “manipulate” search results by stuffing a piece of content with keywords. This bad practice is called keyword stuffing. See this example: 

Example of keyword stuffing.

So if you typed a keyword into Google’s search bar back then, Google would simply show you pages that contained that exact keyword without trying to understand your real intent. This often led to irrelevant results. 

Now, RankBrain can show you pages that do not necessarily contain your keywords  — thanks to its ability to figure out your purpose behind a query.  

Here’s what Google said on Twitter:

Source: Twitter

One way RankBrain operates is by using word vectors. Here’s how it works, as described by Greg Corrado: 

“RankBrain uses artificial intelligence to embed vast amounts of written language into mathematical entities — called vectors — that the computer can understand. If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries.”

For example, if you searched in Google “how to eat a date,” there would be a ton of pages about dates. A date could mean different things meeting socially, the number on a calendar, or a fruit. 

What RankBrain does is that it looks at other words on a page to see if they’re connected to your search query. So if it finds pages with words like fruit, raisin, fresh, dried, or dessert, RankBrain will most likely show those pages to you. 

How to Optimize for RankBrain

As a business owner, marketer, or SEO specialist, how can you create content that’s optimized for RankBrain? Here are some actionable steps to take: 

1. Ditch the one keyword, one page practice. 

Once upon a time, SEOs tried to rank different pages for similar keywords. 

Take, for example, a blog post that targeted the keyword “guitars” and another blog post that targeted “guitar music instrument.” Since these keywords have the same intent, they should be condensed into one article. 

So if you find two or more articles on the same topic on your website, it’s best to combine them into one solid in-depth content. By doing this, you’ll be able to target keyword variations and avoid keyword cannibalization. 

2. Before using a keyword, try to understand its intent. 

RankBrain’s role is to serve content to the right audience. So while you’re researching a keyword, ask: Does the intent behind this keyword meet my audience’s needs? 

Try to put yourself in your audience’s mind. For example, if they wanted to know about the best mirrorless cameras before deciding to buy, wouldn’t they type “best mirrorless camera” into Google’s search bar?

As you can see, organic search results showed product review roundup articles because RankBrain understands that you’re not looking to buy anything specific yet based on your query, making it almost impossible to rank a transactional (product) page for this keyword.  

3. Write content that sounds human.

Gary Illyes said this:

“Optimizing for RankBrain is actually super easy, and it is something we’ve probably been saying for fifteen years now, is – and the recommendation is – to write in natural language.  Try to write content that sounds human.  If you try to write like a machine then RankBrain will just get confused and probably just pushes you back.”

Bottom line? Write in a conversational tone. Like you talk. Keep in mind that most voice search queries use natural language. 

Gary gives practical advice on reading your articles out loud and asking people if they sound conversational: 

“….But if you have a content site, try to read out some of your articles or whatever you wrote, and ask people whether it sounds natural.  If it sounds conversational, if it sounds like natural language that we would use in your day to day life, then sure, you are optimized for RankBrain. If it doesn’t, then you are “un-optimized”.

💡My 2 cents for you: Don’t worry if your article sounds silly at first. You can fix it later on during the editing process.Write for your target audience, not for Search Engines.

4. Know that different queries demand different signals. 

Depending on the user’s query, RankBrain is going to consider certain signals to provide the best content. 

For example, someone searching for the “best movies in 2020” will likely see articles that have been written in the same year. Here, we can see that content freshness plays a big role. Meanwhile, someone searching for “diabetic ketoacidosis” will likely be shown in-depth guides. For this, content comprehensiveness matters. 

That having said, consider the signals that could matter for the search queries you’re trying to rank your content for. 

Final Thoughts 

It’s no secret that most people use Google. We do that because we expect to get the most relevant answers to our queries and RankBrain has made this possible. Knowing how RankBrain works will enable you to take your content creation efforts more seriously. Is your content useful for your audience? Is it relevant? Conversational? 


References used for this article:

1. Bloomberg news story –  https://www.bloomberg.com/news/articles/2015-10-26/google-turning-its-lucrative-web-search-over-to-ai-machines

2. Google RankBrain as part of Hummingbird – https://searchengineland.com/faq-all-about-the-new-google-rankbrain-algorithm-234440

3. Google search statistics – https://www.internetlivestats.com/google-search-statistics/

4. Gary Illyes’ tweet – https://twitter.com/methode/status/879988330074689536

5. Machine learning definition – https://martechtoday.com/how-machine-learning-works-150366

6. Keyword stuffing definition – https://support.google.com/webmasters/answer/66358?hl=en

7. Google SearchLiaison’s tweet – https://twitter.com/searchliaison/status/1108776357880725505

8. Gary Illyes’ RankBrain quote – http://www.thesempost.com/google-how-to-optimize-for-googles-rankbrain/