This is the biggest algorithm update ever made by Google in the past 5 years. The goal is to better understand Internet users’ natural language. If there's one thing Google has observed since its inception, it's that human curiosity has no limits. In fact, 15% of daily searches are new, that is, they have never been typed before. 

This article is based on information from


Google BERT: understanding internet user searches better than ever

Until today when we’re doing research, we type in keywords, one after the other, instead of formulating a normal sentence as if we were talking to someone, in what is called natural language. We do so because we know that there is a greater likelihood that Google will find the results we came for by searching for them with keywords rather than natural language.


Bert, the name’s origin

For several months, Google has been researching neuro linguistic programming. A neural network-based technique called Bidirectional Encoder Representations from Transformers, hence BERT for short.  

Bert, Google’s ‘translator’

This is the result of such research: the words are no longer processed one after the other but one in relation to the other. The aim is to better discern the relationships between words. Google can now contextualize Internet users' requests, and therefore understand their intentions.


What concrete changes for users?

The results offered by Google will be much more relevant. The better the question is understood, the better the answer should be. That goes without saying. The longest requests, and so more natural/conversational ones, with more linking or coordination words are the most concerned. 

As of this publication, only 10% of searches in English are impacted. Deployment to other countries and languages is expected to follow gradually.

Google Bert Update

Updates to the algorithm are frequent, which is why the new information goes against earlier reports from previous months. In fact, on November 19th, 2019, we learned from John Mueller (webmaster trends analyst at Google) that French and Spanish were the next languages that were going to be tested and not before 2020 ("six months to a year").

The launch, which took place in October 2019 in the United States, was apparently followed earlier than expected by other languages. Indeed, on Monday December 9th we learned that Bert was now active in more than 70 languages including French. The expected impact would be similar to that observed in the US.

Examples of changes made by Bert

Here are some concrete examples

        • "2019 Brazil traveler to USA need a visa": Google is now able to locate the word “to” and understand that it’s about a Brazilian wishing to go to the US. In the past (not that long ago! 😛), Google would not have understood the importance of this connection, and would have returned results about American citizens traveling to Brazil, not so relevant :/.


        • "Do estheticians stand a lot at work": interesting case here since, as you know, a word has several meanings (but hey!). Before, Google translated the word “stand” as in “stand alone” whereas now it understands the context and translates it in the sense of holding/supporting the physical effort associated with carrying out the job.


        • "Can you get medicine for someone pharmacy": Google now understands that “for someone” is an important part of this query, whereas before it missed that information and would return general results on medical prescription.


        • "Parking on a hill with no curb": before, negative sentences or including a negation were not Google's best friends 🙁. Now Google understands that you don't want a curb here!


        • "Math practice books for adults"The previous results page would have shown results for a book in the “Young adult” category. Google can now better determine the user’s demographic from the context.



Google results’ relevance is improving thanks to Bert. Bert's effect on the SERP is expected to be substantial, especially for searches using natural language. In addition, voice searches’ boom will consolidate this change on Google results. Supposed/so-called fake or less relevant content, which luckily appeared on Google results due to the poor understanding of the keywords searched by the Internet user, will be deprioritized by this algorithm. And websites with qualitative content will be rewarded for their efforts. That’s why we advise web professionals and editors to continue to write content intended to answer as precisely as possible the Internet users’ needs / questions.

   Article written by Louis Chevant

Further reading

The complete guide to Internal Meshing

The step-by-step method to build your semantic cocoons, your mesh and the optimal tree structure of your website.