As we head into the critical holiday shopping season, Google has begun to roll out what it says is its most important algorithm since the launch of RankBrain in 2014. In fact, the search engine says it’s more significant than that, explaining that its latest ranking technology “represents the biggest leap forward in the past five years, and one of the biggest leaps forward in the history of Search.”
The new algorithm is called BERT (bidirectional encoder representations from transformers) and will affect one out of every 10 searches made on Google. In essence, BERT is a natural language processing brain- transformers are models which process words with consideration to all other words used. Applying these models or transformers to search queries means Google can better understand search intent, therefore returning more relevant results for searches. It will do this by looking at the words before and after in order to gauge what a search user is actually looking for.
Making search more nuanced
BERT will work alongside RankBrain to feed signals back to Google, giving it a better understanding of natural language including nuance and context. This depth of understanding can potentially make a big difference to the search results returned for queries where there is ambiguity. For example, if you wanted to know if a Brazilian visitor to the USA requires a visa in 2019, you may type ‘2019 Brazil traveler to the USA need a visa’. In Google’s own example, it says that pre-BERT it wouldn’t have recognised the importance of to, but post BERT this nuance isn’t lost so the results will direct the traveller to US visa information, not pages for American travellers heading to Brazil.
In addition to reaching more human-levels of language processing in search, BERT will also be applied to rich snippets.
Optimising for BERT
Google’s standard advice is not to change your site or focus on optimising for specific algorithms. As recently as this month it said that marketers and site owners should optimise for users rather than optimise for RankBrain. As BERT builds on RankBrain signals, the same advice applies.
That said, you can optimise your site to ensure that you give BERT reason to return your site in the 1 in 10 cases where search results will be affected. It’s simple to do this too; simply focus on using real, natural language within your page content. There is an opportunity here to review trending and breakout searches on Google to refresh your web site, blogs and other content not just to please BERT but to ensure you’re keeping on top of consumer search habits.
Want to find out more about BERT and how you can use this important algorithm update to drive your optimisation strategies forward? Contact the Adrac team today.