BERT in action! Google applies new ‘BERT’ model to search rankings

After year-long whispers in search engine optimisation (SEO) corridors, Google finally dropped its new search algorithm, BERT. It’s more advanced than the previous language understanding technique because it has the capability to understand language similar to the way humans do. 

BERT was only introduced, and open-sourced, in November last year, but it’s already promising huge breakthroughs for digital businesses looking to get ahead. Google has already declared it as the biggest change to its search system in the last five years, and the overall history of search.

What is BERT? 

BERT, which stands for Bidirectional Encoder Representations from Transformers, is a natural language processing (NLP) technique that is pre-trained using plain text corpus. It’s the first unsupervised language designed to understand search queries better than ever before. It deciphers the full context of a word by looking at the relevancy of the words that come before and after it. 

So what does this mean for SEO? 

To put it simply, this new update will impact 10 percent of all search queries. What that means is, targeted queries will now rank higher than random results on your search. Google spent a pretty decent amount of time and effort testing BERT models to ensure that when you make a search query, your intended result will show up. This is something they’ve been struggling with for years, as the previous search models were not able to accurately provide the intended feedback.

“BERT has helped us grasp the subtle nuances of language that computers don’t quite understand the way humans do,” says the tech giant. For example, when searching “Cape Town flights to Dubai December 2019”, the previous algorithms would typically also return results of flights from Dubai to Cape Town, which is the direct opposite of what the search query was looking for. However, with this new model, you’re getting exactly what’s in your search box. BERT considers words like the “to” in the example above, and its relation to the other words before and after, to understand the query. In this case, BERT identifies that the query is about “going to Dubai” and not “coming from Dubai”. 

Google emphasises that this especially works better in longer and more conversational queries, or searches where prepositions like “for” and “to” matter a lot to the meaning. However, there’s no real way of optimising BERT for search marketing. According to Google's public search liaison, Danny Sullivan, “there's nothing to optimise for with BERT, nor anything for anyone to be rethinking. The fundamentals of us seeking to reward great content remain unchanged.” 

So, yes, we still need to create quality content that has all the information and answers to users’ most burning questions. Like any other SEO company in South Africa, we’re happy to know we can at least optimise content by focusing on long-tail keywords and phrases instead of focusing on just one or two keywords. The second thing BERT will improve is featured snippets. For this, again, the rules of content writing will stay the same if you optimise your content using long-tail keywords and phrases. 

Final take away 

BERT is a powerful NLP technique that’s changing the game for SEO. It’s the first bidirectional tech that tries to understand the context of a word by considering the words before and after. This improves the results of responses for search queries. And, if used correctly, BERT will help content creators improve their ranking and potentially, be featured on Google’s featured snippets. 

For now, BERT only works for searches done in English, but Google is planning a full roll out sometime in the near future. On the other hand, they’re already implementing BERT on featured snippets. Google further explained on their blog that BERT might not solve all search problems, but that it’s all a journey they’re working on. “Language understanding remains an ongoing challenge, and it keeps us motivated to continue to improve search. We’re always getting better and working to find the meaning in-- and most helpful information for-- every query you send our way.”