In SEO, we talk a lot about search intent. One of the keys to success in our business is understanding where customers are in the transactional process. Are they ready to purchase, narrowing their options or just looking for information? Can they be convinced to buy, or is getting them onto your website the challenge in itself?
From Google’s point of view, however, search intent has a slightly more literal definition, which makes it an altogether different hurdle to overcome. In a complex language like English, not only can one word mean several different things – but the words around it can change its meaning further. This means there’s often a gaping chasm between what humans and computers can discern.
Because artificial intelligence is several steps behind us in this area, understanding the true context and importance of every word in a search phrase is something of a holy grail for Google. It’s why SERPs so often display results that aren’t quite what you had in mind.
Imagine you’re looking for a chocolate cake recipe that doesn’t involve using butter. You might type this into Google:
In this case, Google looks at the individual words to find the most relevant results. At the top of the organic results on the SERP it presents you with the result below. It’s searched for the word “no” alongside the words “chocolate cake” (disregarding “butter” entirely) without considering the context.
In October 2019, Google launched what it called “one of the biggest leaps forward in the history of search”. So could its new BERT algorithm bring a long-awaited solution to problems like this?
What is the BERT algorithm?
BERT stands for Bidirectional Encoder Representations from Transformers – but that’s the least important thing about it. In summary, it’s the latest development in Google’s ongoing quest to become more conversational – to better understand search queries and deliver more relevant results. BERT is designed to detect the kinds of linguistic nuances that humans don’t think twice about – to help Google talk the way we do.
The algorithm has begun rolling out across English-language searches, aiming to eventually cover every language Google exists in.
Why is BERT different to algorithms like RankBrain?
BERT’s key differentiating factor is the word bidirectional. This refers to its ability to get context from an entire set of words rather than specifically-ordered ones. Where RankBrain takes a given word and looks at the words directly before or after it, BERT looks at every single one of them.
In theory, BERT should be able to determine the importance of short, adjoining words like “no” in different contexts, as well as keywords. This is what places it a step closer to replicating the natural and irregular flow of human communication. One algorithm doesn’t replace another, so BERT will work in tandem with RankBrain and Google’s other algorithms to improve the relevance of search results.
Will BERT affect every search result on Google?
At present, the algorithm impacts around 10 percent of search queries. While this relatively small proportion is expected to increase, certain search phrases will need BERT’s assistance more than others.
For Google’s artificial intelligence, longer and more involved search queries like “how do I make a chocolate sponge cake with no butter” require a far greater level of discernment than short queries or ones containing brand names, like “chocolate cake” or “Betty Crocker chocolate cake”.
Do I need to optimise for BERT?
In short, no. Trying to second-guess Google’s algorithms or change your SEO approach in response to new developments is never an effective strategy. Danny Sullivan of Google tweeted: “There’s nothing to optimize for with BERT, nor anything for anyone to be rethinking. The fundamentals of us seeking to reward great content remain unchanged”.
If anything, the introduction of a major development like BERT means that we should be optimising less – at least in one sense. We’re a step closer to not having to worry about whether Google will understand what we write – leaving you to focus on writing content that directly answers people’s questions and speaks like they do. SEO has always been about navigating the gap between your customers and your content – and BERT could narrow that gap significantly.