Google’s BERT Update

Background:

Google has revealed its latest AI-powered search algorithm update designed to improve the interpretation of complex search queries and display more relevant search results. The update signals the largest change to search since Google introduced RankBrain (Google’s machine-learning artificial intelligence system) in 2015.

Details and Implications:

Google’s latest AI-powered algorithm changes are down to a new sequence of code called BERT (Bidirectional Encoder Representations from Transformers), a deep learning algorithm related to natural language processing (NLP). It helps a machine to better understand the context of a search query and to more accurately interpret the meaning of the individual words. The company said that one in 10 queries would produce better results.

An example given by Google was for the potential search query ‘parking on a hill with no curb’, the BERT algorithm should understand that the ‘no’ is important in this context and therefore provide more accurate results to the consumer. In its testing phase, BERT has apparently provided far more accurate results than previous algorithm updates.

Vice President of Search at Google, Pandu Nayak, stated: ‘search is about understanding language. It’s our job to figure out what you’re searching for and surface helpful information from the web, no matter how you spell or combine the words in your query’.

The BERT update has initially been rolled out for calculating organic US search results on Google.com. It hasn’t been activated for the organic search results in other languages and countries yet, but it will roll out in the future. As for ‘Featured Snippets’, which appear above the organic search results and include a text, table or list, BERT is already being used for these results in 25 different languages.

Over the past few years, Google has demonstrated the power it has to understand user context and intent in search queries both in Organic and Paid Search. From expanding ‘close variants’ keywords so that they match terms with the same meaning in Paid Search to this latest update, Google continues to leverage its Machine Learning and Artificial Intelligence in NLP to deliver a more relevant and natural experience for users.

This is a huge step towards more accurate search engine results and improving the experience for users. Accurately understanding the context of language is very complex and relies heavily on context, meaning that it may be impossible for Google to ever fully tackle this in its search engine but Ben Gomes, Head of Google Search said that this shift is: ‘something closer to language…a huge step forward’.

Summary:

This is the biggest change in years and Google is future-proofing itself for longer-tail search queries largely being driven by the ease of search on mobile and increased adoption of Voice Search.

For SEO, it’s unlikely that there will be opportunity to optimise towards BERT. Rather, the focus will continue to be on writing quality content and SEO best practice. The change will come for the user and obtaining accurate search results from their query, regardless of their language choice. This is likely to be a welcomed update as driving a more personalised and relevant experience across Search can only help drive more engagement and deliver better business performance.

Further Reading: Search Engine Journal | Search Metrics | AdAge | Google