Google algorithm update: BERT

Language is kind of like an onion: it’s got lots of layers, infinite applications, and can make you cry if you’re not careful.

Search engines like Google are intimately aware of this fact, which is why their algorithms are constantly evolving to better understand the subtle nuances in human language. The better search engines understand these nuances— how you type a query into Google search bar and how you voice a query out loud to your Alexa, Home Assistant or Siri—- the better the search engine can identify your intent, and thus return the most relevant results and featured snippets the internet has to offer.

Enter BERT, Google’s biggest update since 2015’s RankBrain. According to Google, BERT represents “the biggest leap forward in the last five years, and one of the biggest leaps forward in the history of Search.” Google has said that BERT will affect as many as 10% of all new search queries in English in the United States. This is enormously significant for you, the Google searcher, business owner, and/or Search Engine Optimizer. Let’s take a closer look at what BERT does, how it does it, and what it means for the rest of us.

What is BERT?

BERT (Bidirectional Encoder Representations from Transformers) is a neural network-based framework for natural language processing (NLP) in search engines. We realize that probably means nothing to you. So let’s break it down.

Bidirectional- instead of Google reading your search query from left to right, connecting words in a one-directional sequence the way they are read by humans, Google looks at all words in a query (from left-to-right and right-to-left and back again— hence bidirectional) and analyzes their relationships to each other. This is what enables the machine to interpret the intended meaning of phrases that contain words with multiple meanings (like homonyms) and/or context-specific meanings (like prepositions).

Neural networks- systems of algorithms that are designed to enable computers to recognize patterns in data. Computers recognizing patterns in data is the basis of machine-learning— where computers are able to learn things on their own, without human programming.

Natural Language Processing (NLP)- a branch of artificial intelligence (AI) that seeks to enable computers to understand the nuances in natural human language communication.

Natural language

Essentially, Google BERT is a highly advanced AI (Artificial Intelligence) system whose entire purpose is to better understand “natural” language by analyzing the relationships between each word in a sentence or phrase. It’s intended to return more relevant search results for complex, conversational queries. That’s why many have stipulated that Google’s BERT update is aimed partially at voice searches, the kinds you speak out loud to your Alexa, Siri, Google Home and Voice Assistant. For instance, it’s unlikely that you would say “cannabis seo agency” to your home droid, and more unlikely still that you would type “can you tell me where I can find a cannabis seo agency in Portland?” into your phone. No matter how you phrase it, Google will now better understand what you mean with both of those searches.

What is machine learning?

Machine-learning enables a computer to discern patterns in data over time, which allows it to understand the subtleties of homonyms, context-specific and colloquial words.

Prepositions like “to” and “of” can have multiple meanings depending upon the context of the phrase they occur in. With BERT, Google can differentiate the meaning of “to” in sentences like “how to make muffins” and “getting to the Moda Center by bus” by assessing the relation of the word “to” to the other words in the sentence.

What does BERT mean for you, a Google searcher?

What all of this means for you, ever-curious Google user, is that now, when you enter a highly specific, or “long-tail” query into Search (for example, “how does content writing affect your dispensary’s conversion rate?”), Google will better understand what it is that you are asking, and will return more relevant results.

It will read the sentence more like a human would, by paying attention to the modifying words like “how” “does” and “your.” So, instead of focusing on the key phrase “conversion rate” and returning a list of results pertaining to other digital marketing strategies with high conversion rates (like Social Media Marketing), Google will connect the keyphrase “conversion rate” to the rest of the words in the sentence, and will return results that directly pertain to increasing your dispensary’s conversion rate with strong content writing.

What does BERT mean for you, a business owner?

The first thing SEO’s and business owners should know is that there is no way to optimize for BERT on the publisher side. The point of the update is to better understand what searchers mean, to better connect them to information that already exists on your website.

So if you already have a solid content strategy in place, it simply means “keep putting out great content.” Your site will see ranking improvements if it’s full of valuable content (written for humans) that people are curious about. You don’t even have to do anything (besides keep writing).

On the other hand, your site may see a decrease in rankings if it turns out that your content isn’t as relevant to searchers as Google once thought it was (before it understood the complete meaning of the queries that led to your site). So if you’ve noticed a drop in your site’s rankings since BERT rolled out, you may want to do some investigating and hone in on what pages have started to underperform. Refreshing and updating your content is a great way to boost your rankings in general, especially if you do it in a way that addresses the most relevant aspects of your business that a person would be interested in.