BERT – Google’s Latest Algorithm Update
Starting today, October 25th, 2019, Google will begin rolling out a new update to its core search algorithm, referred to as the BERT update. This new Natural Language Processing (NLP) technology will be one of the biggest changes to rankings in the past 5 years. Here’s a look at how BERT works and what it means for your business.
What is Google’s BERT Algorithm?
BERT stands for Bidirectional Encoder Representations From Transformers. This open-sourced neural network-based technique for NLP pre-training was introduced last year and went through months of testing before being released. BERT has made a breakthrough in how Google processes search queries. This model is able to process words in relation to the other words in that sentence, before this, words were processed one-by-one in order. As a result, Google is able to return more relevant results for more specific search queries. The new BERT model can even understand homonyms much easier. For example the sentences “I got money from the bank” or “Meet me at the river bank.” Same word, two completely different intentions. BERT can now easily differentiate the meaning between homonyms and return accurate results.
How BERT Affects Search Result Rankings
The BERT model is being applied to both ranking and featured snippets in Google Search. This is supposed to help consumers find more useful information. According to Google’s Vice President of Search, Pandu Nayak, the BERT model will affect the rankings of 1 in 10 searches in the English language. Most people will not notice the change but this advancement in NLP accuracy is exciting news in itself.
In the examples provided by Google, the results really are that much better, take for example this search query “can you get medicine for someone pharmacy”.
How will Google’s BERT affect my SEO?
According to BuzzFeed News, Google Executives “didn’t say whether websites should expect to see more or less traffic.” They did say, as this update allows Google Search to answer more niche questions, it will hopefully lead to people asking more niche questions. This means that businesses may start seeing increases in their keyword rankings. Some of those uncommon long-tail keywords that Google wasn’t able to understand, will now start returning better results that websites can actually rank for.
What Google’s NLP Update Means for Voice Search
Although Google has yet to explicitly say it, we can assume that the BERT model is getting us one step closer to a voice-first world. The enhancements BERT will make on rankings and featured snippets are only going to increase Google’s ability to answer voice searches accurately. In July, we reported on Google Assistant’s ability to outperform Alexa and Siri with an overwhelming 94% accuracy rate (compared to Alexa’s 10% or Siri’s 12% accuracy scores). This recent update to BERT will only improve Google’s voice search accuracy.
How Brands Can Prepare For The BERT Update
The best thing about this update is that any brand that has already implemented a VEO strategy will be prepared for the BERT update. Brands who are optimizing their content for featured snippets by asking and answering FAQs on their site will be a step ahead of the competition. Making other VEO enhancements such as improving site speed and focusing on more long-tail keywords will have a higher probability of improving their keyword rankings with the new BERT update.
What do you think about Google’s BERT update? Have you seen any changes in your own searches as a consumer? Let us know in the comments below.