BERT: Google’s Algorithm Update One Year On
- A year since the Google Algorithm Update and BERT is still developing to improve human and search engine connectivity.
- The Google BERT update stands for Bidirectional Encoder Representations from Transformer.
- BERT has been rolled out across 25 different languages.
- BERT uses natural language processing to better understand how humans are thinking and how this relates to search engine searches.
- Voice command is increasing the demand for natural language processing and BERT is set to develop
- DeepRanking is a search ranking incorporation of BERT to improve search results accuracy in response to the specific wording of search questions.
A year after BERTs full search engine Google update, we’ve come to love this strangely named update which caused quite the initial stir within the industry. After a year in action, BERT now affects 10% of search queries – a massive number when you consider that Google process 3.5 billion searches per day.
So, what have we learnt about BERT over the last year, and if you’re incredibly late to the party what is all this BERT talk about?
As a full-service digital marketing agency, our experts are well versed in the world of SEO and the changes that come with a Google algorithm update, and BERT is no exception. If you want to tap into our SEO specialist knowledge, and ensure your website stays visible, get in touch today on 0800 088 6000.
Table of Contents
- What is BERT?
- What is the BERT update?
- How Does BERT Affect Search Quires?
- How Does BERT Affect SEO?
- Where is BERT progressing to?
- BERT and Google DeepRank
What is BERT?
The Google BERT update stands for Bidirectional Encoder Representations from Transformer. It was designed to help Google better understand what people type into their search bar, especially when it comes to conversational search.
Conversational search starts a dialogue from the search, asking the user click-based questions to guide the users to what they’re interested in and, this is one of the key aspects of the BERT update.
The BERT update was released on the 24th October 2019 but it has been in development since early 2018. Since then, BERT has been rolled out across 25 different languages.
Unlock the Power of SEOCall Our Experts Today
What is the BERT update?
Google created an in-depth algorithm analysis to further understand the questions search engine users, and customers, are asking their search engines to unlock the different meanings behind language. It was essentially an update to help Google understand the nuances of dialect and natural language processing.
Ultimately, Google want to better understand how humans are thinking and how this relates to search engine searches.
So, what is natural language processing and why is it so important for digital marketing? Natural language processing is a model created by web developers to help computers understand our language and get over the communication barriers.
How Does BERT Affect Search Queries?
Just some of the exciting search developments rolled out by Google to better understand our search habits:
- Pre-training from unlabelled text – BERT is the first NLP framework that is pre-trained on pure plain text instead of on labelled corpora. This training was conducted through unsupervised learning.
- Bi-directional contextual models – True contextual understanding is only possible when you can see all the words in a sentence at the same time — giving you the understanding of how every word impacts the other words in the sentence. As a machine, Google has historically struggled with this context — until now.
- Transformer Architecture – Reinforcing the point above, the use of transformer architecture allows Google to analyse each word in relation to the other words in a search query. Without this, the query could have an ambiguous meaning.
- Masked Language Modelling – By design, the BERT architecture will analyse sentences that have some words randomly masked out. It then attempts to predict what the “hidden” word is correctly, and factor that into its determination of the search query intent.
- Textual entailment (next sentence prediction) – One of the most exciting features of BERT is its ability to predict what you will say (or type) next.
Stay Ahead of Search Engine UpdatesFind Out More Here
How does BERT Affect SEO?
Firstly, don’t panic – there’s no need to change the way you do SEO for the BERT update. The main principles still apply and keyword research and quality content will ensure your SEO stays strong. However, there are some slight shifts to what BERT deems important:
1. Informational Queries
Google wants to provide it’s users with quick, direct and relevant answers to users questions. If you can clearly answer a question in your content you’re more likely to be features by Google in a feature snippet to answer these queries
2. Keyword Focus
Leading from the previous point, BERT has shifted keyword focus to favour long-tail keywords and questions to give readers a direct answer and understanding. Often, long-tail keywords can broaden the search conversation and reach a larger audience, however this doesn’t negate the important of short keywords. The best course of action is to create a keyword strategy focused on a blend of both long and short keywords.
3. Longer Content Isn’t Always Better
BERT tracks user engagement and time spent reading your content, so you need to tailor your content to specifically target the search or query and provide enough content to be useful and relevant, but not drown the reader so that the key answer is lost.
4. Dynamic Search
PPC advertising should focus on dynamic search ads to target relevant search quires based on your website content. This will leverage Google’s intelligence to help generate more consumer data and work together with BERT to help you drive ROI on your campaign.
5. Adopt a Customer Centric Approach
The BERT update is geared towards improving the users search experience and providing them with relevant, useful and trusted content to give people information and results they want to find. As a creator, you need to focus on delivering highly-focused content to please the BERT update and improve overall user response.
Take the Hassle Out of SEOCall Our Team Today
Where is BERT progressing to?
Everything is set for Google, and other search engines, to be constructing bigger and better variants of BERT to build on the strengths of BERT. Especially incorporating the growing use of voice search and voice-enabled devices.
‘Hey Google’ and ‘Alexa’ voice command searches are on the rise. Gen Z are an emerging generation of customers and clients who increasingly use voice enabled search. As this generation ages, the use of voice generated search will age with them.
This creates a demand in the technology of search, teaching Google to become smarter and integrating an understanding of conversational language and regional dialect to understand the users and deliver the expected results. This is an incredibly challenging and exciting development for search engines and brands alike.
BERT and Google DeepRank
Following a year of the BERT algorithm update, Google released a documentary on how search works, specifically looking at deep learning and ranking.
The video details that DeepRank was launched in 2019 as BERT, and works with BERT’s learning methods to rank aspects of search. We can think of DeepRank as BERT’s interface in Google Search, the two are integrated and are working towards the same aim of further understanding human search methods.
DeepRank builds on Google’s current machine learning and natural language understanding. Search ranking has progressed a lot in the past two decades, but prior to BERT’s natural language progressions, search ranking couldn’t understand the subtleties of language in the same way humans do. Enter DeepRank – a BERT integrated ranking system which has the power to understand human language terms.
DeepRank allows BERT’s knowledge to be applied to search ranking and ensures that relevant and useful search results are produced for each search query. This requires a massive computer power and countless hours of testing to allow DeepRank to understand the intricacies of natural language and conversational search. This proves that the BERT update is only just getting started with all the improvements it can make in search, and our search engines are set to increase in intelligence and human understanding over the next decade.
Build Your Unique SEO Strategy NowGet In Touch
To tap into our SEO specialist knowledge and make the most of all of Google’s updates, get in touch today on 0800 088 6000.