What is Google RankBrain? The definitive guide

Human engineers still work on the algorithm, of course. But today, RankBrain also does its thing in the background.

In short, RankBrain adjusts the algorithm by itself.

Depending on the keyword, RankBrain will increase or decrease the importance of backlinks, content update, content length, domain authority, etc.

It then analyses how Google searchers interact with the new search results. If users like the new algorithm better, it stays. If not, RankBrain reverts to the old algorithm.

Here’s the crazy part:

Google asked a group of Google engineers to identify the best page for a given search. They also asked RankBrain.

And RankBrain outperformed Google’s clever brains by 10%!

In short, RankBrain works. And it’s here to stay.

Now that you’ve seen an overview of what RankBrain is, let’s dig deeper into how it works.

How RankBrain works

RankBrain has two main jobs:

1. Understand search queries (keywords).
2. Measure how people interact with the results (user satisfaction).

Let’s look at each of these.

How RankBrain understands any keyword you search for

A few years ago, Google had a problem:

15% of the keywords people typed into Google were never seen before.

15% may not sound like much. But when you process billions of searches per day, that equals 450 million keywords that baffle Google every day.

Before RankBrain, Google would scan pages to see if they contained the exact keyword someone was searching for.

But because these keywords were new, Google had no idea what the search engine really wanted. So they guessed.

For example, let’s say you searched for “the grey console developed by Sony”. Google would search for pages containing the terms “grey”, “console”, “developed” and “Sony”.

Today, RankBrain really understands what you’re asking. And it provides a 100% accurate set of results:

Not bad.

What changed? Before, Google tried to match words in your search query to words on a page.

Today, RankBrain tries to understand what you mean. You know, like a human would.

How? By matching never-before-seen keywords with keywords that Google has seen before.

For example, Google RankBrain may have noticed that a lot of people search for “grey console developed by Nintendo”.

And they’ve learned that people searching for “grey console developed by Nintendo” want to see a set of results about game consoles.

So when someone searches for “grey console developed by Sony”, RankBrain shows results similar to the keyword they already know (“grey console developed by Nintendo”).

It then shows results about consoles. In this case, the PlayStation.

Another example: a while ago, Google published a blog post about how they are using machine learning to better understand searcher intent:

In that post they describe a technology called “Word2vec” that converts keywords into concepts.

For example, Google says that this technology “understands that Paris and France are related in the same way as Berlin and Germany (capital and country), and not in the same way as Madrid and Italy”.

Although this publication did not specifically refer to RankBrain, it is likely that RankBrain uses similar technology.

In short: Google RankBrain goes beyond simple keyword matching. It converts your search term into concepts … and tries to find pages that cover that concept.

In chapter 3, I’ll show you how this changes the way we should do SEO keyword research. But first, let’s cover the most interesting element of what RankBrain does …

How RankBrain measures user satisfaction

Sure, RankBrain can try to understand new keywords. And it can even modify the algorithm itself.

But the big question is:

Once RankBrain shows a set of results, how do you know if they’re really good?

Well, you watch:

In other words, RankBrain shows you a set of search results that they think you’ll like. If a lot of people like a particular page in the results, they’ll give that page a boost in the rankings.

And if you hate it? They’ll drop that page and replace it with a different page. And the next time someone searches for that keyword (or a similar term), they’ll see how it works.

What exactly does RankBrain observe?

It pays close attention to how you interact with search results.

Específicamente, está mirando:

• Organic click-through rate
• Dwell time
• Bounce rate
• Pogo-sticking

These are known as user experience signals (UX signals).

Let’s look at an example:

You pulled a muscle in your back playing tennis. So you search for “pull back muscle” on Google.

Like most people, you click on the first result. Unfortunately, the introduction is full of fluff and filler content (“Your back is an important muscle group…”).

Then you hit the back button and check the second result:

This one isn’t much better. It’s full of generic advice like “rest and freeze your back”.

Then you hit the back button on your browser one more time and give result #3 a try.

Keyword research in a RankBrain world

As you saw, Google can now understand the INTENT behind a keyword.

So, does that mean that traditional keyword research is dead?

No!

That said, you may need to adjust your keyword research process to be more RankBrain friendly.

Here’s how:

Ignore long tail keywords (they are obsolete)

Long tail keywords are dead.

In the past, it made sense to create hundreds of different pages … each optimised around a different keyword.

For example, I would create one page optimised for “the best keyword research tool”. And another optimised for “best keyword research tool”.

And good old Google would rank each of them for their respective long tail keywords.

Today, RankBrain understands that these terms are basically the same. So they show almost identical search results.

In short, optimising for long tail keywords no longer makes sense.

What should you do instead? Read on…

Optimise around medium tail keywords

Instead of long tails, I recommend optimising content around medium tail keywords.

Medium tail keywords are intermediate terms. They get more search volume than your average long tail. But they’re also not incredibly competitive.

For example, here’s a set of keywords on the topic of “Paleo Diet”. The terms in the middle are mid-tail keywords.

When you optimise your page around a mid-tail keyword (and make that page awesome), RankBrain will automatically rank you for that term… and thousands of similar keywords.

In short, I recommend optimising your page around a single keyword.

(Just make sure it’s a mid-tail keyword).

Then, let RankBrain rank your individual page for many different related keywords.

How about an example of this process in action?

Example of keyword research and on-page SEO in a RankBrain world