Google announced the most significant jump in the last five years and make it one of the biggest leaps in search history earlier throughout the week. That is a BERT algorithm update, a way to better understand one of Google’s few queries so people can understand them.
I covered it in detail in Search Engine Land, but I summarize it in bullet point fashion for those quick readers:
- BERT started rolling out earlier this week and will be fully live by the end of this week. It is live for queries in the American English language.
- It also widely affects snippets, and snippets are depicted with not only the original search for the English language but also several different languages.
- Here means that it is probably the fluctuations that we saw this weekend and the middle of the week that were related to this.
- BERT is very much like RankBrain in that it is a machine learning algorithm that aims to understand page queries and content better.
- Technically, BERT is a pre-training method focused on neural networks for natural language processing (NLP). BERT means the bidirectional encoder representative of the transformer. Google wrote more deeply about it here last year.
- BERT enables Google to understand human-related queries, more natural language-based queries, and content. BERT helps Google understand the nuances and context of search terms and better match queries that are more relevant.
- Google says it is enormous, by and large, it affects 10 percent of all searches, but Google also says that it is the most significant search move in the last five years, and it is overall the most in the search history. It is one of the big stages.
- We saw changes this weekend and the middle of the week and thus, it seems towards SEO.
- Google said that they tested BERT extensively and the company is seeing significant benefits for using it.
- RankBrain or other language algorithms are not replaced; it can be used in conjunction with them.
- I think you can’t optimize for BERT because you can’t optimize for RankBrain – just write for people.
- Neural matching is also different.
- It is not as relevant as core updates or other updates – at least still from the SEO point of view.