31
Oct

The Google BERT Update

“What do you mean?” This is Google’s ultimate and fundamental question. They want to answer your question. They want to provide the resource you need. They want to know what you really mean. The Google BERT update began the week of 28 October 2019 and is intended to answer that question.

Welcome “BERT”

The acronym stands for “Bidirectional Encoder Representations from Transformers”. This update from Google is about understanding human language. It began to roll out last week. It uses A.I. (artificial intelligence) on a neural network powered by machine learning. Google already knows perfect grammar, spelling, and punctuation, but it needs to know lingo, intention, and meaning. That is what BERT studies and interprets.

“Bass” can mean the lowest male singing voice or a fish.
“Rose” can mean a flower, a color, a wine, or a cut.
BERT uses context to understand meaning.

The Impact of Google BERT

Instead of just trying to match keywords from the search to a website, BERT is trying to match your intent to a web result that gives the most relevant answer.

This is a big change! Google has said it is their biggest update of the last five years and will impact 10% of all searches in the English language. (Other major updates affected 2% – 5%.) The goal is to understand the meaning of longer, conversational queries such as:
“cost of a college degree in a job interview”
“credit report needed for car financing”
“2019 brazil traveler to usa need a visa”
“sale commission to real estate agent”
“do estheticians stand a lot at work”

This will impact:

Your SEO and BERT

Nathan Greenberg runs our SEO strategies and in his opinion BERT is going to negatively impact websites that employ “keyword stuffing”. He also believes it is going to impact image and video search as more ADA-compliant content is created utilizing ALT tags. For Arkside clients, no strategy change is necessary. We have not observed any negative impact on any client websites at this point from Google BERT!

Bottom line: It reinforces the need to write website copy for human readers instead of search engine bots.