What is the BERT update?
BERT (short for Bidirectional Encoder Representations from Transformers) is the latest update from Google which puts context into search queries. It is essentially an extension of RankBrain; a machine learning-based Google algorithm update that launched in 2015. Rankbrain incorporates the use of natural language processing to make the search engine more conversational in the way that they are trying to get the user’s question answered.
The BERT update goes a step further by trying to understand the context and understanding that language is more than just text on a page. The biggest change here is Google is incorporating an understanding of the use of prepositions (to, for, etc.) in order to better understand the context of a search query. This update comes as no surprise to SEO experts (including us) and signifies a notable step towards improving the user experience on all devices, including voice search.
Rather than looking at words within a search term in silos, BERT is able to consider the full context of a word by looking at the words that surround it. This means that search results will become much more relevant as Google will be able to understand the full search query rather than individual words.
Ultimately, the purpose of BERT is to improve user experience by better understanding the contextual meaning and the search intent behind every search query. Google’s search engine is a product and users are the customers. Like any business, Google is trying to improve its product by cutting down on poor quality content to ensure it can serve highly relevant results.
What does BERT mean for websites?
For websites, this means that quality content is more important than ever if they want to see continued improvement in their SEO. Google and users don’t want the ‘marketing fluff’ that fills so many sites. Gone are the days of using “keyword density” for one single keyword on a page. Now Google is looking for more in-depth content that uses long-tail keywords, descriptors, and modifiers. Take from the teachings of your college English professor and write your webpages and blog content explicitly. “Show, don’t tell.”
While Google always says “write for users, not the bots” we believe you can do both. Things like incorporating more long-tail keywords to increase topical relevance and adding a descriptive header structure are loved by both users and bots. By using AI-powered software, LEWIS is able to optimise pages in the most effective way possible, without keyword stuffing or overcomplicated layouts. At LEWIS, MarketMuse is one of our favourite tools we use for expanding on existing content to help build out our topic relevancy. We’re able to analyse the top-performing content in Google for a specific query and score our content against that. MarketMuse then incorporates deep learning and natural language processing (similar to Google’s RankBrain algorithm) to tell us where to fill in the gaps by using relevant long-tail keywords. By approaching on-page optimization this way, we are able to give both users and bots exactly what they need.
In addition to Google trying to understand the contextual intention behind a search query, Google released an E-A-T update which received less of a fanfare, but still deserves attention. E-A-T (Expertise, Authoritativeness, and Trustworthiness) is a key factor in quality rankings and should be considered in the development and optimisation of all site pages.
Google is quite clear in its search quality evaluator guidelines that pages need to demonstrate E-A-T throughout and have a clear purpose. This advice is still incredibly important, even more so now that BERT has arrived. Google is slowly starting to piece together and better understand the intention behind every webpage, so making sure each webpage offers answers to your target audience’s questions will be crucial for garnering rankings and driving organic traffic to your site.
Pages that have thin content and low code to text ratios are another common error that causes sites to rank lower than they would otherwise are thin pages. Having a large number of pages that are low on content and lacking word count and detail could really slow your SEO progress down. Why? Google doesn’t think you can successfully answer a question in 300 words. Studies have shown that content with 1k words or more performs better in search. However, with the rise of quick answer boxes and featured rich snippets, we might start to see a shift. Especially with voice search on the rise, quick short answers might end up being more ideal for the user. With voice search on the rise and the contextual update of BERT, adding schema markup to your webpages will shift to be a priority for SEO in 2020 and beyond.
Ultimately, BERT simply points to what SEO experts already know – user experience is crucial. Through both design and copy, user questions need to be answered, E-A-T needs to be demonstrated, and a clear understanding of target personas and answering their questions and solving their problems.
What is the real impact of the BERT Update?
The good news is that the world of SEO isn’t too different from before BERT. Sites won’t be immediately penalised like they were when Penguin was introduced, however, if your content isn’t contextually relevant your site might not perform as well as it could.
The update does, however, present a key opportunity for companies to put their sites under the microscope and take a closer look at how they can improve and optimise their content for BERT and RankBrain. Using tools such as Google Analytics, SEMrush and MarketMuse, you can analyse your content against the competition, and optimise your pages for topical relevance. Check poor behaviour metrics and any page two keyword rankings to spot opportunities for improving existing content.
BERT won’t cause sites to immediately lose rankings, but by actively optimising site pages you will start to see growth an improvement. Regularly re-visiting pages and improving them with E-A-T and tools such as MarketMuse should prevent your site from becoming stagnant, and instead, start to improve your SERP rankings.
Three things you should review on your site
Ultimately, BERT is a content-focused core update and it comes as no surprise to SEO experts. At LEWIS, we’ve been actively optimizing client websites for some time which means that BERT should have a positive impact on our sites.
If you haven’t been actively optimising don’t worry, it isn’t too late. Here are three things you should review on your site following the launch of BERT:
- Evaluate the copy on the page – contextual relevance is crucial in BERT, so take advantage of AI tools and take a second look at your copy.
- Explicitly answer user questions – remember that BERT is all about quality. This means your content needs to clearly answer user questions.
- Make use of longtail keywords – this will help you add more context to your page and help move you away from any internal terminology that your users might not understand.