Google BERT: Google’s Biggest Update in 5 Years.
Of Searches Affected
According to Google, the BERT Google algorithm update is the most significant search engine change of the last 5 years. Designed to better understand language searches, BERT is expected to impact 1 in 10 search queries. For now, BERT only impacts english-based searches but as their technology evolves, expect BERT to support other languages in the future.
Google has remained consistent with their messaging regarding algorithm updates, strive to improve search results to fit user’s goals and align searches with the most relevant and useful results. For that reason, it’s important to stay abreast of these Google algorithm updates and continue to make the corresponding improvements to your website both from a content creation and an SEO perspective.
BERT stands for Bidirectional Encoder Representations from Transformers. To clarify, this is not Michael Bay’s new robot movie. It is a neural network-based technique for natural language processing (NLP). The BERT Google NLP is a uses a machine learning algorithm designed to better understand searches and internet content. What that means is as time goes on, BERT will become more refined and accurate with delivering Google search results.
Algorithm updates in the past Google NLP BERT considers a query in its entirety, using the content and relations of all words in the search, rather than on-by-one in order. This enables BERT to figure out the full context of a word and the intention of a search by looking at the relationship of each word. This is where the ‘bi-directional’ part of Bidirectional Encoder Representations from Transformers comes from.
With BERT, Google can better understand what a user is searching, even if it is disjointed or if there are misspellings. There’s no more need for ‘Google speak’ or rephrasing google searches to get better results. Using longer, more conversation queries, Google can now provide more accurate results with pages optimized for Google Bert (more on that later).
Google expects that 15% of all future search queries are completely unique – they have never occurred before. To do so, Google is employing open sourced AI/Machine Learning coupled with improved supercomputing processor speeds to more relevant results, faster. Using BERT Google AI, trending, viral, and news-related searches will now provide more improved and accurate results.
A good example of this comes from Search Engine Journal. The search phrase was “how to catch a cow fishing?” Have you ever heard for cow fishing before? Well, in New England, a large striped bass is often called a “cow.” Before, Google wouldn’t know how to decipher this dialectic nickname. With BERT, Google now uses the word “fishing” to add context to the remaining search query – providing accurate results for bass fishing rather than farm animals.
BERT is also designed, as you’d expect, to help with voice and text search using the Google Assistant (Okay, Google). According to Wordstream, by 2020, 50% of all searches will be voice searches and 30% will be done without a screen.
Mobile voice searches are 3 times more likely to be local based rather than text searches. Also, Amazon Alexa searches tripled during last year’s holiday season, compared to the previous year. This means that local search and e-commerce results will be hugely impacted by Google BERT.
Although only 15% of searches are expected to be impacted with Google BERT, the overall trend, along with smaller, more frequent algorithm updates mean that specific SEO tactics can be employed to further improve rankings. Here’s what we can expect for BERT and future algorithm updates: