Background to BERT
When search engine giants Google announced that its 2019 BERT update would be its biggest in over 5 years and that it would impact around 10% of all search queries everyone involved in SEO stood up, listened and probably if they are being honest gave out a little shudder – a result of a strange mix of terror, apprehension, excitement and unease. Big updates like this can get into the psyche of the most stoic of SEOs and while they can manifest in some changes that must surely be acted upon they can also act as a scapegoat for any shortcomings an SEO strategy might already have. Let’s dig a bit deeper, first looking at the details of the BERT update in and of itself. Firstly it’d be good to get a handle on BERT itself. BERT stands for Bidirectional Encoder Representations from Transformers and is a complex deep learning algorithm that has to do with natural language process. Its primary function is to aid in the ability of a machine to comprehend the meaning of words in a sentence taking into account all of the potential nuances of context. So, this is the system that was updated. And that’s why SEOs were all of a quiver.
However, when Google explained that the BERT update would be focusing on pretty long tail search queries (more than a few words) you could almost hear the sighs of relief in the SEO community, given that they are far less focussed on such search terms. But perhaps, given that such a huge effort has been made to update the accuracy relating to them, long tail searches will increase in prominence in SEO strategy as a result. What it will certainly mean, in terms of rankings – in the simplest language – is that for around 10% of such searches of a few words or more, the rankings will better represent the user’s intention than before. This can only be a good thing for SEOs and users as long as the former pay great attention to the most popular searches of the latter when it comes to their own business sector and tailor their content to it.
How BERT affects SEO and rankings
While BERT and its update don’t affect rankings in an immediate and direct way; because it relates to text and how well that text is understood by the search engine, there is a definite potential for a long-term knock-on effect on SEO strategy and potentially in the rankings of a page down the line. Those complaining that this particular update had a profound short term impact on their rankings are, however probably mistaken. It was not an update to Google’s algorithm for ranking, rather to its in-built brain’s ability to correctly interpret long tail search terms.
While it is probably true that at present there are no new tools or strategies that SEOs can implement that actually target the revised BERT effect, what almost certainly holds true is that the update has made it even more important for web pages and their content to accurately represent the products and services on offer and perhaps more importantly, the searches that see users land on their pages.
Identifying and optimising the META user search queries is probably the biggest, single most important thing that SEOs can do not only in the light of the update, but in the light of the direction Google and other search engines have been headed for some time. This will also be the best way web managers can try to focus on the sort of descriptions and explanations that can see them being highlighted in Google’s much envied ‘featured snippets’ sections.