Google’s BERT Algorithm Update And How It Affects Your Website
On October 25th, 2019, Google rolled out what they are calling “the biggest leap forward in search in the last five years”. This update aims to help Google understand searches better than ever before and is already affecting roughly 10% of all searches. We’re going to help you understand what exactly this update means and how it might affect your website’s SEO strategy.
What Exactly Is BERT?
In Google’s words, BERT is a neural network-based technique for natural language processing called Bidirectional Encoder Representations from Transformers, or as Google calls it–BERT, for short.
In plain terms, BERT is a new technique Google is using in its search engine to more clearly understand and process searches that use every day, natural-sounding language. More specifically, it helps Google provide much more specific answers to questions on search results, even longer phrased questions with subtle context.
How It Affects Your Website
Right now this update ONLY affects Google’s featured answer shown in search results. So technically, this update affects search and not your website. However, Google pulls the content it shows in featured results from your website, so many websites will notice positive and negative outcomes from this update.
Here are some examples of searches before and after the BERT update:
Before: For the search “do estheticians stand a lot a work”, Google matched the keywords “esthetician” and “work”, pulling up an article that ranks well and includes both of those keywords in the title. Seemingly unable to understand the context of the question.
After: Google now understands the context of the question, and is able to show a specific answer even though the keywords in the title don’t match as well. Google is able to look past this using natural language processing, resulting in a much more relevant answer.
Before: For the search “Can you get medicine for someone pharmacy”, Google shows a featured answer that closely matches the keywords used in the search, but is not able to understand the context of the question.
After: Google now understands the question is about picking up a prescription for someone else and is able to show a much more specific and relevant answer to this question.
For more examples of searches before and after the BERT update, check out Google’s article here.
The End Of “keyword-ese” Is Approaching
So why has Google decided to roll this update out? Just like any other algorithm update, they are aiming to make search more useful to users. Google has recognized that the future of search depends on being able to respond to natural language, not what they call “keyword-ese”.
We have all searched in keyword-ese before, this is when you search for something like “Edmonton Dentist” instead of something more natural like “Who is the best dentist in Edmonton for fillings?”. Google realized that search results we’re not able to properly answer longer questions with more subtle context and needed to update their ability to understand these more complex searches.
Bert’s Current Limitations And Future Possibilities
Currently, the only thing holding Google back from using this update for more searches is their ability to process the data. Google’s servers currently don’t have the hardware to roll this functionality out for all of search. Although, this could quickly change considering Google announced a major achievement in quantum computing two days prior to rolling out BERT.
The future possibilities of BERT are vast, but in regards to SEO, this is an opportunity to optimize your website for questions specific to what your users are searching for. Right now, if you can give a specific answer to a specific question, there is a better chance you will be able to show in Google’s coveted featured answer placement, which could help you get more quality traffic to your website.
Thinking ahead into the future, it’s never been more important to start using natural language on your website. This involves focusing on using phrases and sentences that read naturally, instead of simply trying to match keywords on your pages that users are searching for. As Google has always recommended, provide useful and relevant content for your users, not for Google.