Google Update BERT
The acronym BERT stands for „Bidirectional Encoder Representations from Transformers“.
BERT is based on Natural Language Processing (NLP) and neural networks and should be able to better understand language. The update should understand the relation to all words in a search query and not consider a sentence or a question word for word as before, but also the words before and after. BERT can thus look at the full context of a word and help the user to make their requests in a natural way.
Why an update
According to Google, search queries are getting more and more similar to real, human communication, and they are getting longer and longer. Already, around 70% of search queries are assigned to the long-tail sector. The user wants to have a precise answer from Google to now whole question sets.
Currently, BERT is already active for organic search results and featured snippets. Featured snippets already use the update for two dozen countries. So it is also working to identify the German Featured Snippets, while the Update to identify Organic Search Results is currently helping to better understand one in ten US searches in English. So BERT is not yet active for the German search results. Only over time is BERT to be rolled out for other country indices.
Google also benefits from BERT in the area of voice search, as verbal requests are usually provided in complete sentences and the algorithm can help to better understand and classify these complex search queries.
Great content is rewarded by Google. Because BERT aims to better understand human language, it is important to write texts as naturally as possible and to create them to match the needs and intentions of users.
Simply knowing what is most searched for to optimize it is not enough. It is important to understand the users, what the different motives for an interaction represent, how close you can stay to the user and shorten the customer journey. User Intent and user experience count.
* Rank Brain
This algortihm was implemented by Google 2015 and, like BERT, is based on human language. However, RankBrain uses machine learning to better understand search queries. BERT will not replace RankBrain but complete it.