Google BERT is a new search algorithm. How will the ranking change and what to do now?

Officially: Google introduced a new search algorithm Google BERT.
What information is available at the moment?
How does the search algorithm work? What will change in search engine optimization?
We will deal with the questions below.

Google BERT

What is Google BERT? Google BERT is a search engine algorithm designed to improve the relevance of search results by the ability to analyze not key phrases, but sentences.

To achieve the goal, a neural network is used, which is designed to give the search engine an understanding of the text in different languages. BERT stands for Bidirectional Encoder Representations from Transformers or Bidirectional Neural Network Encoder .

Where is the algorithm used? At the moment, the algorithm is being tested in the next issue – United States of America, English. The technology is used on 10% of search engine queries. In the future, this algorithm will be used for SERP in most countries and languages.

What is changing? Previously, Google analyzed each query as a set of key phrases , and selected relevant results using the analysis of key phrases. With BERT, the search engine will understand the entire context of the request. Such an opportunity is achieved due to the ability to analyze not only key phrases, but also auxiliary words in search queries (for example, phrases and prepositions).

So, without Google BERT the search engine will analyze key phrases. With the use of the Google BERT neural network, the goal of understanding sentences will be achieved.

Changes will also affect snippets, as the technology will be used to form fragments in search results.


How to optimize the site for Google BERT? At the moment, the only recommendation is to write texts for people and expand the semantic core of the page. To attract more traffic will be able to pages on which:

  • Content is hosted with information architecture ;
  • Content covers the needs of the intent .

What makes sense in terms of information architecture:

  • Design the information space of the site so that the design facilitates the completion of tasks and intuitive access to content;
  • To structure the information that is presented on the pages of the site.

What makes sense on the pages in terms of improving the intent :

  • Expand pages with text, with additional relevant keyword phrases from search visibility. Search visibility is uploaded through special services;
  • Analyze search queries for which traffic to the site was attracted and add relevant phrases to the content;
  • Add relevant phrases from Google, Bing, and Yandex search suggestions to the content. This source contains low-frequency phrases that are not found in other systems;
  • Analyze internal search logs and add relevant relevant key phrases to the content.

How to unload site key phrases from search visibility? The easiest way is to unload data from the MegaIndex database.

The robot regularly scans search results, collects and updates lists of key phrases by which sites are ranked.


Google started using the BERT ranking algorithm. The announcement of the new algorithm is posted on the official blog.

Google calls the BERT algorithm the biggest breakthrough in the last five years and one of the biggest achievements in search history.

This technology is aimed at solving problems on natural language processing , in other words, we are talking about tasks on the topic of natural language processing.

In general, this technology falls into the outline of the Google AI First.

By the way, Google BERT is a completely open technology, which means there is data to study.

The algorithm affects:

  • Ranking part of pages in search results;
  • Formation of snippets.

How to prepare for the arrival of the algorithm? It makes sense to expand the list of key phrases with similar ones. As sources for downloading such phrases should be considered:

  • Search engine on the site;
  • Analytics system;
  • Search suggestions;
  • Search Visibility.

Alternatively, key phrases from search visibility can be downloaded using the service to search for similar key phrases.

What good is Google BERT for search engine optimizers? When creating title and h1, a dilemma often arises: choose title for SEO or title for CTR ? With the introduction of Google BERT for a number of pages, it will become possible to adapt the title to CTR requirements, without compromising search engine optimization. But you need to test.

What good is Google BERT for users? Getting relevant search results will be easier. Using voice search will produce more relevant results.

Adapted for Israel. Original

  • More info