Understanding Google’s BERT Update
This past November, Google rolled out a significant change—one that’s been billed as its largest algorithm update in half a decade. The last update of this magnitude, known as RankBrain, was confirmed in October 2015. The update introduced a new ranking factor that was aimed at better understanding users’ true intent by using machine learning.
Now, Google is expanding upon that machine learning to bring us BERT, a deep learning algorithm that will allow the search engine to better recognize natural pattens of speech.
If all this is starting to sound a bit uninteresting and unhelpful to you, trust me you’re not alone.
But let me tell you why you should care.
BERT will affect an estimated 1 in 10 search queries—not an insignificant number, I’d say—so if you haven’t already, now’s the time to get familiar with the update that’s changing the way Google search operates.
What is BERT?
BERT stands for Bidirectional Encoder Representations from Transformers. (I’m sure instantly clears everything up, right?)
Basically, it’s a deep learning algorithm that was designed to aid in natural language processing by putting more emphasis on the nuances of context. It’s meant to better align the algorithm’s interpretation and understanding with how people actually speak.
Here’s how Google explains it:
These improvements are oriented around improving language understanding, particularly for more natural language/conversational queries, as BERT is able to help Search better understand the nuance and context of words in Searches and better match those queries with helpful results. Particularly for longer, more conversational queries, or searches where prepositions like “for” and “to” matter a lot to the meaning, Search will be able to understand the context of the words in your query. You can search in a way that feels natural for you.
With the rise of voice search, this improvement in language processing is more important than ever for Google if it wants to deliver relevant results to users posing questions in a more natural, conversational way.
How Does BERT Work?
Let’s take a look at an example provided by Google. Here, the user is asking if it’s possible to pick up someone else’s medicine at the pharmacy.
Prior to the BERT update, Google’s algorithm did not understand the meaning of the prepositional phrase “for someone,” so that context is effectively scrubbed from the search results. After the update, however, the search results are more relevant because the algorithm has been designed to recognize the context of the preposition “for.”
Here’s another example of how Google’s BERT system more accurately delivers relevant results.
With this search query, you and I can clearly understand that the user, a Brazilian traveler, is trying to get a visa to the U.S. However, without understanding the context of the word “to” Google’s old algorithm mistakenly directed the searcher to a result for U.S. citizens traveling to Brazil.
With a better understanding of these contextual words and phrases, the update makes it easier for the algorithm to pick up on the complexities of natural human speech patterns.
However, this is just the beginning. Even Google admits that this update is a jumping-off point for more research and experimentation into improved language processing. It’s not suddenly impossible to stump the algorithm, but it is harder than it’s even been before.
How Do I Prepare for BERT?
Contrary to popular belief, you don’t really have to do anything to prepare or optimize your site for BERT. The point of this update is simply to help the algorithm better understand the search queries, not penalize sites that don’t adhere to certain rules.
If your site has been negatively affected by BERT, chances are that you weren’t supposed to be getting that traffic in the first place. It’s possible that the algorithm was previously leading searchers to results that were not relevant to their actual intent, and you were reaping the benefits of that misunderstanding. That traffic likely wasn’t doing you any good from a conversion standpoint anyway, so don’t worry too much if your site traffic took a small hit in the last couple of months.
As always, your best bet to attract more traffic to your site is by creating high-quality, long-form content. That much hasn’t changed, and it’s unlikely to do so as Google continues to tweak and improve its language processing abilities.