Google Bert Update: What changes in SEO?
On October the 25th 2019, big SEO community sites like the SEJ ( Search Engine Journal ) posted that Google anounced its new Algorithmic Update, the so called Google BERT Update. After that many sites from the SEO community started mentioning about the Google Bert Update while they tried to guess what will change in Google Search and what impact will have the Bert Update.
Two months after the Bert Update l am in the position to assure you that nothing changed for the websites that rank in Google. And today the sun will rise from the east, the kids will go to school and the small cattles will continue to eat grass on the prairies.
But, let’s see what exactly changed for Google with Google Bert Update.
What is Google Bert Update;
A small research l made on Google, l found that BERT is the acronym of the phrase Bidirectional Encoder Representations from Transformers.
BERT is a neural network-based technique for natural language processing (NLP) pre-training. About a year and something, Google delegated to a very sophisticated and searchwise guy, Pandu Nayak which is the Vice Prisedent of Search to improve the Google answering machine and make it more precise.
l know that in this post maybe l will use very technical terminology, but l will make an effort to keep the basic things as simplest as it can get and l will try to to do what every big Poet from the Homer ages do, to introduce big concepts in small things like this post.
Google had become into an answering machine from a search machine it was. l thing that everyone had noticed that under the search field Google displays on the position 0 some relevant search terms or search questions with the searched keyword.
The reason why is very simple. According to Pandu Nayak, the last 15 years, the 15% of the total searches are completely new search terms that had never been searched before. Out of billions of searches the 15% are total new searches.
These new searches are even more complicated from the already known ones and consist of key phrases with more than three words that demand even more precise answers.
How to change a spark plug in a gas running car
How to recover my Netflix password etc.
Searches now are becoming more and more complex and the previous technology of the Google Algorithm could not answer with precision because it couldn’t undestand and predict very well
the real user intent behind every search.
There was a clear need for the use of a more sophisticated artificial intelligence system that would simulate human thinking with the use of machine learning.
The process of training an artificial intelligence computer is simply called training and the training is done mechanically by the system itself. That is, the artificial intelligence computer learns by itself what the real internet user is looking for today by tracking which search results he clicks and adjusts the search results accordingly.
The scientific definition of Google Bert Update
In 2018, Google announced a natural Transformers class, NLP pre-training, codenamed BERT. This technology allows everyone because it is open source to train their own system of Q&A.
The Transformers category is several algorthmic models (software programmes) that process the written word over the other words in the sentence, rather than reading them one by one. BERT models can better understand the overall text of a paragraph rather than looking at the words before and after, that is, trying to understand the intent of the user in a Google search.
Google has therefore introduced this innovation into its algorithm and significantly improved its search results initially for English search terms and for some other languages such as Portuguese.
She then announced that she would introduce BERT in all national languages and this will impact every national search engine of Google like Google.fr, Google.de etc.
In order to implement this very ambitious plan, it needed to upgrade its existing supercomputing infrastructure and use the latest technology (expensive technology) in Cloud TPU due to high processing power requirements
This is the well-known Google Cloud that has been upgraded with new resources and new machines that are open to the public and to you. At the near future we would definitely all buy one to have it in our living room and we will show off saying, do you know what TPU I bought from Google?
The conclusion is that Google theoretically and practically made a leap forward in artificial intelligence technology, invested millions of dollars in new infrastructure and you’re trying to outsmart her with 100€ SEO per month?
BERT in action
Google itself through Pandu Nayak has released several tests that prove that Google BERT was more accurate and clearer in searches than before. No one else has been able to perform such tests except Google itself because everyone does not have access to its search algorithm to open it and close it as he would like.
Below are some of Google’s tests.
2019 brazil traveler to usa need a visa
In this Google search: “2019 Brazilian US Traveler Needs Visa”, our friend BERT (you will see below why I say so and how much he was trolled by the SEO community on social networks) managed to understand that the intention is that somebody from Brazil wants to go to America and get a visa, not the other way around.
In its pre-BERT search, Google had messed it up, showing exactly the opposite of the search results that an American wanted a visa to travel to Brazil.
Do estheticians stand a lot at work
In the next search phrase: “Do estheticians stand a lot at work” (which means: Beauticians are standing up for their job) Google gave a response that was related to the job of beauticians as a job, completely ignoring the physical requirements of the job that is to make the aesthetic stand up to their work.
But now with Bert realizing the true nature of the question has finally quessed right from Google and so Google gave the right answer.
Another success of the BERT model.
Successful search is not a problem solved
According to Google again, its purpose has not yet been achieved in order to reach perfection in searches and to provide the most relevant results. But with the introduction of BERT into the core of its algorithm, a big improvement step was made that brought it closer to what the average user was really looking for, literally guessing more than what it was actually looking for.
If it was a novel it would have the title: The search behind the search
Of course, we have to add that BERT is constantly learning and evolving, so in the future the improvements will be more and more noticeable to everyone.
Google BERT huge troll in social media
Google BERT Update has welcomed the crazy SEO community by featuring him with the puppet of the Muppet Show, Bert!
I think Bert and the Muppet show was a very successful franchise that broadcasted at a global level on the 70s and 80s.
Here’s a little snapshot of Bert dancing with a fiery blonde, supposed to be the personification of Google?
And the trolling didn’t stop there, dozens of tweets were posted by accounts of SEO experts, SEO agencies and various other websites such as:
How to train BERT
Which is a parable of Google’s expression that the system is open source and anyone can use BERT to train their own response engine.
That is, Bert can also enter your site as a bot to respond to your visitors or clients in chat.
-Hello Bert, here’s President Trump
The Impact of Google Bert Update on SEO
Let’s not forget, we are an SEO company and so we should see this new change of Google’s algorithm from the perspective and perspective of SEO.
n terms of increasing or decreasing organic searches it does not change anything significantly, although Google claims that Google Bert Update affects 10% of searches which are mainly more specialized queries rather than simple keywords.
The new upgrade, however, favors web sites that have a question-and-answer structure and answer the most important questions of your prospective customers.
You should fill out those that do not have FAQ-type texts and posts within your landing pages and enrich them with new content that users are now looking for.
At the same time because Bert was trained on Wikipedia, it is advisable to organize your website based on the structure of Wikipedia, with many internal and external references, internal links and external links i.e. images, videos and large posts with more than 2.000 words or more.
Don’t be afraid to exceed the 100 links per page that Google has imposed as an artificial limit to avoid being considered spam. Wikipedia itself, in its dozens of entries, has many more links without ever being penalized. So it’s all relative on SEO and it’s all about a more holistic approach to the subject.
If you want to improve your website based on Bert Update and increase your searches on Google, feel free to request a website promotion offer today!
Want to get on the first page of Google?
If you have a website and want to rank it you may ask for a quote, we are doing this job for more than 7 years and our team of SEO experts can find the best solution for you.