Share This Article
Each so often, Google is making a major change to its main algorithm. This was the launch of BERT in October 2019.
But with hundreds of Google updates per year, what sets BERT apart from the rest of it?
Ok, according to Google, BERT is the biggest change in five years (since RankBrain) and affects one in ten search queries.
Put simply: BERT promises to answer the questions you have on Google more specifically.
What is Google BERT?
BERT stands for Transformers’ Bidirectional Encoder Representations.
While this sounds complicated, what we need to know is that it’s Google’s way to better understand the finer points of the natural language we use.
In reality, before Google implemented BERT as part of its key algorithm, it was published as an open-source neural network.
It meant that someone with technological know-how could use the code to develop their own state-of-the-art answering device.
Yeah, that sounds like a lot of fun, but how does BERT work?
How Does Google BERT work?
In the past, Google has not always been successful at interpreting nuanced or conversational searches.
And according to Google, “people sometimes type strings of words that they think we’re going to understand, but they’re not really how they would normally ask a question.”
This trend is known as “keyword-ese” in Google HQ, and it’s something Google needs to stop.
As a consequence, BERT extracts the complete meaning of a word from a question through machine learning and natural language processing. This then discusses the words that come before and after it, rather than only concentrating on one word at a time.
Google did a lot of testing before releasing BERT into the wild. Check to see if the tests before and after have been changed. Below are two examples of the assessments to be reviewed:
The preposition “to” is necessary for the keyword “2019 brazil traveller to use a visa.”
This shows that the traveller needs to visit the USA from Brazil, not the other way around.
Until BERT, Google did not appreciate this complexity or the full meaning of the question and would advertise results for US travellers travelling to Brazil, which, of course, is not what the searcher wants!
For the term “parking on a hill without a roadblock,” Google would have previously been puzzled by this question. It would place too much focus on the keyword “curb” and ignore the word “no.” Again, it would produce results that weren’t exactly what the consumer wants.
From Keyword Matching to Keyword Intent Matching
Today, determining what makes reliable and appropriate content is something that Google has been working on for a while.
In 2013, Google updated its key algorithm and named it Hummingbird.
After that, Google’s determination of whether or not a page was deemed important was primarily related to how well the keyword was configured.
When you use keywords in essential on-page elements such as title tags, headings, and in the content’s layout, you will have a high score. You can think of this method as 1.0.
One of Hummingbird’s main objectives, however, was to move away from “strings, to things” by better understanding the meaning of keywords and how they relate to other subjects, rather than the string of characters that make up a word.
Fast forward a few years ( 2015) and Google launched Hummingbird’s first new sub-algorithm: RankBrain.
It’s been remarkable since it was Google’s first artificial intelligence algorithm.
RankBrain better understands the user ‘s purpose behind a keyword and focuses on achieving the best possible outcome rather than the best-optimized keyword outcome.
Google has gone public in saying that RankBrain is one of its top three rating factors.
So, here’s what you need to know … there’s a growing change away from just a keyword matching a search query to fulfil the underlying keyword purpose of a question.
It makes a lot of sense if you think about it. Google is now using artificial intelligence to try and figure out whether or not a consumer is happy. Keyword purpose matching is rapidly becoming one of the most critical aspects of modern SEO.
So, what is the difference between BERT and RankBrain?
Okay, let ‘s continue with some similarities. They are powered by Google’s main Hummingbird algorithm, use artificial intelligence, and concentrate on understanding the meaning behind a search query, so they can help produce the best possible result.
RankBrain excels in interpreting the user ‘s purpose for long-tailed, more descriptive and unknown keywords, while BERT discusses the finer aspects of natural language.
Once you start thinking about the user intent behind the keyword and what the best outcomes might be for the user, you have advanced your thought from the 1.0 keyword match to the 2.0 keyword match.
Would you need to change your SEO approach?
It’s probably not.
Google has been telling, for what seems like an eternity, “focus on the consumer and everything else is going to follow.”
Google’s approach hasn’t changed, however with advances in artificial intelligence, Google is now becoming better at knowing what the consumer needs and working harder to recognize the consumer ‘s motives behind even more keywords.
Modern SEO now incorporates both relevance 1.0 (keyword matching) and relevance 2.0 (keyword purpose matching) to generate the appropriate content. All of which are central to your SEO performance.
Today, with the implementation of artificial intelligence through Hummingbird, RankBrain and now BERT, it’s obvious that Google is doubling down to provide its users with the best user experience.
Are you doing the same thing?