Meet BERT, Google’s new algorithm that enhances search results by understanding the context of your queries and delivering more relevant results.
Ieva Treiliha
BERT is Google’s latest algorithm update, and it may well transform the SEO landscape as we know it. BERT marks a significant step forward for Google in better understanding users’ search queries.
BERT was first introduced last year, and BERT stands for Bidirectional Encoder Representations from Transformers.
This means that BERT can interpret the correct meaning of a word by looking at the words before and after it. If all goes according to Google’s plan, BERT will lead to a better understanding of search queries compared to the current model, where each word is processed in sequence.
In this article, you can learn more about:
At present, BERT is still in a testing phase. So far, Google uses various BERT models to better understand search queries. However, over time, BERT will impact both standard search results and featured snippets.
Currently, BERT is not used for 100% of searches. At the moment, BERT is only applied to 1 in 10 searches in the USA in English. Google states that the BERT algorithm is so complex that it pushes the limits of Google’s hardware, which is likely why it is only used for a limited number of queries.
Before you completely overhaul your SEO strategy to get ahead when BERT arrives in Denmark, remember that BERT analyses search queries – not websites.
That said, as it stands right now, the BERT update is likely to have a significant impact on your on-page SEO. SEO experts will increasingly need to consider user search intent when working on on-page SEO. Half-hearted on-page content will not thrive with Google’s BERT update.
Google has tested BERT and various versions of this algorithm and has provided a few visual examples of how a search result may appear before and after BERT. You can read the full article here.
In this example, we can see that before the BERT algorithm, the search result displayed an article stating that US citizens do not need a visa when travelling to Brazil. However, there is a crucial error here, as the query is actually asking for information about a visa from Brazil to the USA.
This demonstrates how BERT can improve search results, giving users a more accurate answer to their query because of HOW BERT reads search queries.
Previously, Google matched individual words in the search query with words in the search results. In the example above, Google matches the phrase “stand-alone” at the bottom of the first result with the word “stand” in the query. But in this context, that is not the correct usage of “stand”. In the second result, we can see that BERT understands that “stand” is related to the physical demands of a job and therefore shows a more relevant answer.
In this example, Google struggles to decode the query due to the word “no”. The system focuses so much on “parking on a hill” and the word “curb” that it ends up showing an irrelevant answer.
BERT, on the other hand, focuses on all the words in the query and can spot the subtle nuances that can make a world of difference to the result presented to the user.
As has been mentioned many times before, best practice is to focus more on users’ search intent rather than keyword density in your on-page text. It used to be a popular SEO hack to use the primary keyword as many times as possible in the text in order to secure top rankings.
This technique is outdated. Today, we focus much more on our readers and potential customers. There are three types of keywords people typically use, depending on their search intent:
From the examples we have seen so far, BERT will primarily affect top-funnel keywords, the so-called “informational keywords”. If you aim to maintain your high rankings for these and perhaps even get ahead of your competitors, you should really focus on different variations of relevant informational keywords. Read more about keyword types and search intent here.
The key is to provide users with a clear answer to their question, rather than writing a 2,000-word blog post stuffed with exact-match keywords.
Google has neither confirmed nor denied it, but there are SEO experts who speculate that BERT will support one of Google’s recent major updates – E-A-T, which stands for Expertise, Authority, Trustworthiness.
Google is made up of algorithms and systems in several layers. Google constantly updates existing algorithms and adds new ones to improve their product. The aim is to give users the best and most precise answer to their search query.
E-A-T focuses on content quality and provides guidelines for qualitative content, while BERT is a language algorithm. So, in principle, BERT and E-A-T have very different functions, but share a common goal.
It is a mistake to assume that Google’s updates are independent and separate, when in reality they work together. There may be thousands of algorithmic layers working together to improve the search results users see after performing a search query. Read more about Google algorithm updates here.
There is a great deal of confusion and click-baiting around BERT at the moment. People in the SEO industry have many speculations about what BERT will mean for our daily SEO work, and BERT has become a very popular topic in various SEO groups online – but it is not always clear what impact BERT will actually have. Here are some of the most common myths about BERT that I have seen discussed:
BERT does not mean your website MUST be optimised for long-tail keywords. BERT’s primary task is to better understand users’ search intent and then adapt the search results accordingly.
So, BERT’s main job is to understand users better, and for that reason, it is wise to keep your readers in mind when creating new content for your website. Google has actually recommended as best practice for some time now that you think of the user before the search engines. There is a good chance that a version of BERT will arrive in Denmark in 2020, so it is important to consider BERT in your SEO strategy already.
Some in the SEO industry believe this algorithm is not particularly important. However, Google itself has stated that BERT will affect every tenth search result in the USA – and that is just to begin with. 10% of all search results is significant if BERT has the same impact on other languages.
Right now, it is difficult to assess how much BERT will affect the Danish SEO landscape. We do not even know when the BERT algorithm will be implemented in Denmark.
The assumption that BERT is not a major update may stem from speculation that BERT will only affect long-tail keywords and other longer search phrases that are easily misunderstood. Some believe that traffic will be directed to the correct landing pages when it comes to long-tail phrases, but that BERT will not have much effect when it comes to exact-match keywords. At this point, however, nothing can be said with 100% certainty about the algorithm. To draw firm conclusions about BERT, we need more time.
Right now, you can find countless articles, forum discussions, videos and other materials about BERT. Everyone has an opinion and wants to be ahead of the curve when BERT is fully implemented, but it is important to remain critical of some of these speculations. Opinions are divided: Some believe BERT has the potential to revolutionise SEO, and others downplay its significance.
Google itself has stated that BERT is the biggest update in the last five years, and it is not often Google makes such statements – in fact, they usually keep their cards very close to their chest.
At WeMarket, we believe that BERT has great potential to change the SEO landscape, and it is important to stay updated on BERT’s development while maintaining a critical approach to your sources.
If you have any questions about the BERT algorithm, or need help with SEO in general, you can contact us here.
If you would like to read more, here are some of the latest blog posts from WeMarket:
Effective synergy between SEO and SEA
Schema Markup – Get more visible on Google
At WeMarket, we offer businesses a benchmark report that compares their marketing efforts with their key competitors. You decide which competitors we should compare against.
We specialise in selling physical goods online and growing webshops – and now you can benefit from this expertise, even if you’re not already a client.
It’s completely free.
