Just when you think you have a handle on all the factors that impact search engine results, and you’ve put a solid SEO plan in place, something changes. Since you’re here reading this article, we are going to assume that you have been hearing about this mysterious thing called BERT over the last couple of weeks. Let’s clear up the confusion and talk about what BERT is and what it means for your business.
What is BERT?
BERT is an acronym for Bidirectional Encoder Representations from Transformers, which is a bit of a mouthful and not very self-explanatory for most of us, so let’s break it down further.
To really understand what BERT is, it helps to start with what BERT does.
What does BERT do?
The purpose of BERT is to improve the Google search algorithm’s skills in natural language processing (NLP), or accurately understanding human language. It does this by changing the way the search algorithm looks at words.
NLP algorithms like BERT are called pre-training representations. They control the framework and methodology that the search algorithm uses to understand the meaning of words. NLP representations can be designed to be context-free or contextual.
A context-free representation has a single understanding of the meaning of a word. For example, if your search included the word “tears,” the results that came back could include either sites about how to mend a tear in your jeans or sites about watery eyes and depression. The other words that surround the word “tears” in your query aren’t considered. Clearly this is the least advanced model and is non-optimal.
A contextual representation looks at surrounding words for clues on how to interpret your query. There are two types of contextual representations – unidirectional and bidirectional.
A unidirectional representation looks at the words before the word in question and uses them to get context. If you searched “how to stop tears in my nylons,” a unidirectional representation would look at “how to stop” for clues on the contextual meaning of the word “tears.” That’s an improvement on the context-free model, but in this query, the search algorithm could still think you are looking for allergy medicine for watery eyes or support for sadness.
The ultimate type of pre-training representation is a bidirectional contextual representation like BERT. When BERT considers your query “how to stop tears in my nylons,” it considers text in both directions to build a deeper basis for contextual clues. Since it considers both “how to stop” and “in my nylons,” BERT will better understand the context of how you are using the word “tears.”
How will BERT impact search results?
As BERT helps Google’s search algorithm to understand the intricacies and nuances of more conversational long form queries, the search results that are returned will become more accurate.
This is good news for both the person doing the searching and for you as a business. With BERT, if your business appears in a search result, odds are it is because you offer the type of answers, products or services that a potential client is looking for. You are relevant to their needs, so they are more likely to engage with your content and less likely to bounce because it wasn’t what they were looking for.
You may see a slight decrease in the number of times you appear in search results, but it should be offset by the higher percentage of engagement on the search traffic you do receive.
How should I change my SEO techniques to align with BERT?
Knowing a little more about what BERT is and what it does, you may be wondering if there is anything you should do to tweak your SEO techniques. Good news – if you already have a good SEO strategy in place, just keep doing what you’re doing. Avoid keyword stuffing and keep creating high quality organic content rooted in solid keywords. In fact, those long tail keywords that you have developed will pay off now more than ever because BERT will allow Search to better understand their meaning.
If you aren’t familiar with long tail keywords and you are looking for some guidance on figuring out how to get started with SEO, simply reach out to us and we would be glad to give you some pointers.