Google’s BERT algorithm is a significant upgrade and enhances their understanding of language in order to get better search results. It is probably the best step forward in search since RankBrain. Considering that BERT will have an impact on 1 in 10 search queries, this guide is something that no webmaster can ignore. The BERT update is one of the biggest changes that Google has made in search and it is important to optimize for the BERT algorithm so that your site does not lose its ranking in search queries. After BERT was rolled out, there have been some mixed reactions. Some website owners have reported that they lost 30-90% of their organic search traffic while others claimed that it had no impact on their site. This should not be underestimated because if 30% of traffic is lost, it can affect the revenue of a website, essentially being the difference between running a business and not. The impact of BERT was such that Google’s John Mueller shared tips on how to recover lost traffic after BERT during a webmaster hangout. This just goes on to show how important optimizing for BERT really is.
What is Google’s BERT Algorithm?
Google’s BERT algorithm was introduced in late 2019, and since then it has played an increasingly important role in how Google ranks websites. BERT stands for Bidirectional Encoder Representations from Transformers, and honoring its name, it signifies that the algorithm observes and seeks to make sense of words in relation to all other words in a given sentence. As a different to past algorithms, BERT can consider the entire context of a word by taking a look at the words that come before and after it. So, why is this important? Well, it’s been suggested that the change has helped Google to understand intent behind search queries better, and has also helped the search engine to comprehend user’s queries in a more natural way. Consider the past search query “2019 brazil traveler to usa need a visa”, someone utilizing a broken English search would be more likely to have their query misunderstood under past algorithms. Google would have tried to match the keywords of the search to keywords on different pages in hopes of finding a relevant result. With BERT, Google is now capable of understanding that the user wants to know if a Brazilian person can travel to the US without a visa. This pivotal change in understanding user intent has a large impact on the results and is something all web-owners should take into account.
Importance of Optimizing for BERT
In a recent interview with Pandu Nayak, he introduced MUM (Multitask Unified Model) to explain the concept. He said that suppose there’s a word “affect” which has two different meanings. Firstly, it’s about one thing having an impact on the next: cause-and-effect. Contrast this with affect, which may be one of the most common words in English, but one of the toughest for people to agree on a meaning. Here, affect is used to indicate the approximation of feeling or emotions in BERT, which changes the meaning in the sentence. BERT will be able to interpret all the words around and check to bring the right meaning of affect.
Targeting more on the message, BERT is the most recent update from Google. Google said it’s one of the biggest updates. The main thing to consider is that it’s not a penalty. Most of the sites that appear to be in a miserable condition after this update were not actually penalized. So what’s the real matter? Their ranking drops might be due to irrelevant content. Google never made changes for local languages in previous updates, but now with BERT, Google is able to understand the nuances and context of words in English, and then you get more local language.
Overview of SEO and BERT
If you are an SEO professional or site owner focused on optimizing your website, then you would have probably come across the significance of the BERT algorithm update. When it comes to search engines and optimizing for them, most of us focus on understanding how to rank in the image pack, video carousel, news results, and importantly the organic search results. The click-through rate from a ranking position in any of these areas in the SERP to the destination site often determines the ROI from the effort to rank there. However, the recent BERT update and the subsequent natural language understanding AI models that Google will develop have made it necessary for SEOs to understand the way Google now understands queries and content on your website. This is an oversimplification of the machine learning and NLP revolution regarding search, but I hope to spell out what it means and what you should do.
Understanding BERT’s Impact on SEO
This has an interesting effect on SEO. In recent times, there has been a trend of articles with long-winded titles that clearly state the article is about a certain topic. This is an attempt to target a specific search query. For example, there may be an article titled “The Impact of Brexit on UK businesses and the potential economic effects in the future.” In the past, this article may have been written without this specific title in an attempt to more naturally explain the topic. But the writer may have felt the need to make sure the title targeted the search query and, in turn, felt the need to include specific keywords throughout the article. This could mean that the title was poorly constructed just for SEO purposes and the article may not have even touched upon the points that the title suggests. Now with BERT, Google will understand the context of the article regardless of the title and may bring itself more relevance to search results of Brexit-related topics, without even needing to specifically target such a narrow search query. So, less focus on specific keywords means less keyword targeting and more of a need to write high-quality content that will now be more relevant to search queries.
When it comes to understanding BERT’s impact on SEO, it is important to understand how the algorithm understands and processes search queries. Google is hoping to give more relevant results to searchers and, in turn, this has spurred more search engine optimization to better understand user intent. BERT changes the way search queries are processed by Google. Before BERT, there was a much greater focus on specific keywords in a search query. For example, if someone were to search “math on how to change a tyre on a car,” the main focus may have been the words “math” and “tyre,” due to these being seen as the keywords. The search results may not have been that relevant to the user. But BERT will consider the whole search query and understand the context of it, thus providing more relevant search results. This means that there is less of a need to target specific keywords as Google will recognize the relevance of an article to a search query, even if the article does not specifically target those keywords.
How BERT Affects Search Queries
Search queries often have a great deal of nuance that is lost on search engines, leading to incorrect assumptions about what information the user is really after. For example, how might a search engine interpret the query “2019 brazil traveler to usa need a visa”, shown below? Prior to BERT, Google’s algorithms viewed the words “to” and “need” as unrelated. As a result, the main focus of the query was on the traveler needing a visa to the USA. But the word “to” signals that the traveler is asking if a Brazilian needs a visa to travel to the USA. The addition of BERT helps Search better understand the context of misspelled words, too. “Would a Brazilian need a visa to go to USA” has a mistake in the word “Brazilian” (a common misspelling). In showing the importance of the word “a” in this context, the old algorithms did not understand that the user was asking if a citizen of Brazil needs a visa to travel to the USA, and not the other way around.
BERT’s Focus on Contextual Understanding
This represents a major change in the way Google is interpreting search queries and has already had a major impact on the search results. To understand this, we can compare BERT to a human translator. Hearing the sentence “I just got a job at Google” is quite ambiguous, as it’s not clear whether the person is expressing that they previously worked at Google and have now found a new job or if they are saying that they have just landed a job at Google. The context of the sentence depends on what was happening around the time of speaking, and BERT is aiming to understand this in the same way that a human would. In the example above, a search for “its the same job at google” fill form is now more likely to return results relating to information on how to fill out a form to apply for a job at Google, as BERT recognizes that the word “job” in the searcher’s query is related to the search “I just got a job at Google”, even though the two queries are from different contexts. This has significant implications for on-page optimization and also keyword research, as it means that Google is now able to better understand what a piece of content is about and also the context in which certain keywords are used.
Implications for Keyword Research and Optimization
Optimizing your website for Google’s BERT is very crucial because BERT lays the foundation for change in the way SEO works. In the past, SEO service was mainly about finding and using the right keywords. Then, the focus shifted into creating as much high-quality content as possible to rank high in Google’s search results. The major change here is Google’s shift from being an information provider to an answer engine. This means that instead of providing users with links to sites that might have the answer, Google now aims to provide the actual answers to the queries. BERT’s ability to understand the context of the content on your website makes it easier to match your content with an answer to a user’s query. This means that the better BERT can understand the content on your website, the more likely it is that Google will select your site for the direct answer to a query.
BERT’s Effect on Featured Snippets
Featured snippets are a great way to use your content to draw users further down the SERP to your site. They are a great visual for your website, showing that your content is relevant to the search query. However, with Google’s BERT update, there is concern that featured snippets may take the content we want to get users to click on away from the SERP and to the snippet. This is because BERT is making the SERP more functional. This means that it is providing searchers with the information they are looking for without having to click anywhere. An example of this would be two years ago, you would search the question “what year did ww2 end” and would likely click into a website to find the answer. Nowadays, Google provides you with the information at the top of the SERP, thus taking the traffic away from websites with the information. This is the concerning nature of BERT for featured snippets. A prime example of which is conveyed in the tweet below. This presents us with a slight issue. Good quality content should retain a high CTR. If users are getting the information they need from the snippet and not clicking through to your content, then it is not providing a good ROI. This is causing people to question whether to optimize for featured snippet now provides the same value as it did previous to BERT.
Strategies for Optimizing Your Website for BERT
Creating High-Quality, Contextual Content
Next, make sure that your content has a clear and focused purpose. While it’s important for your site to have a wide breadth of information, each page should have a specific topic and target a specific keyword. Pages that are about a single topic are more likely to rank better in search, and your audience will be more likely to find the information they’re looking for.
The first step is to conduct a thorough content audit. Understand what’s on your site and whether it’s still relevant (or ever was) to what your website is all about. Ask yourself, is this something that people would look for and find useful? If not, it might be time to rethink the purpose of the piece of content. You don’t necessarily have to get rid of it, try moving it to a different location on your site if the information is still relevant to something.
For the first strategy to improve your website’s performance with BERT, creating high-quality, contextual content is the key. With a better understanding of natural language, BERT is able to relate to the nuances in meaning in language and serve better results. There is no special wizardry to creating quality content, this is something you most likely hear often. And that’s the truth. By focusing on content quality, you are investing in your site’s findability, relevance, and credibility. Here are some tips to get you started.
Utilizing Natural Language and Long-Tail Keywords
Incorporating relevant, well-written long-tail keywords into your site’s content can help to improve your website’s relevancy and strengthen the potential for search traffic. Long-tail keywords are 3-4 word phrases that are very specific to what your site offers. An example of this would be a short-tail keyword such as “dog food” whereas a long-tail keyword would be “organic wet dog food”. Long-tail keywords are beneficial to website content as they can allow a greater understanding of the audience’s need to be established and can break down the content into specific areas. Long-tail keywords are particularly important when considering the types of queries a user would be likely to voice search. This is essential in today’s market as 20% of all searches on Google are now voice based and are normally employed in situations where the user needs to use their mobile phone for information, providing directions and location based information. Due to the precise nature of long-tail keywords it is advised that content written specifically for these keywords are not changed. For websites with existing good quality content, optimizing site content for natural language search queries with the use of long-tail keywords is often relatively straightforward and simply requires the addition of long-tail keywords into the content. However, for new websites or web pages with low content, it may be useful to construct content around specific long-tail keywords, tailoring content to the audience and offering comprehensive information specific to their needs.
Improving User Experience and Site Speed
Google claims that BERT does not bring significant changes in terms of technical optimization. Nevertheless, this doesn’t mean you should ignore the foundations of enabling Google to read your website easily. Using a fast, reliable web host and managing a light, clutter-free website theme are easy ways to improve speed. If you’re using WordPress, changing options in settings and installing plugins may require further reading if you are unfamiliar. In terms of an easy and effective solution, plugins such as WP-Optimize or W3 Total Cache offer a simple, one-click solution to improving WordPress speed. If you do not feel comfortable with any of this, consulting a web developer is a great idea. This is the cleanliness and optimization of a website’s code, improving loading times and site speed. This does hold importance, as a slow website or pages that do not fully load are likely to be skipped over when Google is indexing sites and retrieving information.
Improving your website’s user experience and site speed is pivotal to successfully optimizing for the BERT algorithm or any content strategy. Although this does not include a specific part of the algorithm, this lays a critical foundation for the next steps to optimization. BERT is seeking to better understand the nuances of language and match that with search queries. The closer the language on your website is match towards search queries in terms of context, the more Google will favor your site in those results. Similarly, improving site speed will make it easier for Google to index your site and retrieve information in a quick and efficient manner.
Enhancing Structured Data and Schema Markup
Schema markup is a type of structured data that is a semantic vocabulary of tags (or microdata) that you can add to your HTML to improve the way search engines read and represent your page in SERPs. Search engines need this added information to better understand the content and provide the best search results – giving you an advantage over websites without markup. This can also be done using the Google Structured Data Markup Helper.
When considering structured data on your site, you should use the new BERT related itemprop to give the search engine more context about your content. This will help the search engine better understand your page and can be used in the future to help enable special search result features. Be sure to test your markup using the Rich Result Test to ensure that it is free from errors and is working as intended.
Monitoring and Analyzing BERT Performance
Website optimization is a continuous and ongoing process, and monitoring and analyzing BERT performance with respect to the previous state is crucial to make sure the website is heading towards the expected direction. It is very important to check individual pages periodically to see if they are still relevant and aligned with business goals. Make sure to monitor search traffic for each important page, the query the page is ranking for, and the average position. This will give an indication whether the page is getting more or less visibility in search results. In order to understand the implications of BERT on search traffic and user behavior, it is good to consider using controlled experimentation (A/B testing) to understand the impact of BERT changes. This will typically involve comparing two versions of some content where only one version is changed. A random sample of users are then shown either the original or changed version, and their behavior is compared. This type of experiment may reveal changes in user behavior due to BERT algorithm changes, which may be difficult to discern from comparing site level data before and after BERT.
In addition to using more natural language on your website, increase your semantic signal by using structured data, which is a great way to help search engines further understand what your content is about. Don’t miss this opportunity to take advantage of an easier way to inform search engines and drive more organic traffic. Although there is no guaranteed way to get rich results, adding structured data to your site can help search engines better understand your content and in turn provide better visibility in search results. Now looking at how search result appearance is affected by BERT, it is even more crucial to ensure that your content is the best that it can be. With featured snippets, paragraphs, lists and tables, there are many ways that your content can be pulled to best answer a user’s query. Make sure that you have interesting and high-quality content useful to searchers and in doing so you can possibly increase the chances of having your content being pulled as a rich result. This will hopefully result in higher CTR and more traffic to your site.
Recap of BERT Optimization Strategies
Write like you talk. This is the most powerful way to optimize for BERT. Instead of optimizing for keywords, write content that is useful and comprehensive. This is how you should be writing for your site’s audience anyhow. Think about the questions that your readers have and answer them. How can you best serve your readers with your content? What is your content trying to accomplish? Content should not be simply about having some words on a page; it should be about communication. Phrase your topic in the form of a question and simply answer it. Read your content aloud. If it sounds stiff and formal, rewrite it. Language that is unnatural to you will also be unnatural to your readers and to BERT!
The strategies for optimizing your website for Google BERT are elegant in their simplicity: Write like you talk; Use clear and succinct language; and consider the user’s global language.
Future Considerations for SEO and BERT
With BERT, the search algorithm has taken a huge step forward in understanding the finer nuances of user intent and contextual meaning within a search query. This is crucial to understanding the future of on-site content, and how it will be optimized for search engines. For years, SEOs have been writing content and linking based on topic relevancy and not taking user intent into consideration – this is how we have ended up with such a huge proportion of low-quality, nonsensical articles clogging up the search results. These articles may well be ‘optimized’ based on the keyword they are targeting, but quite often they fail to serve the user’s true intent. New Internal Linking Strategies An effective way of guiding Google towards understanding the context of a given page is to revise internal linking strategies. In the past, SEOs have been telling Google what a page is about by using descriptive anchor text which is heavily targeted on a specific keyword. This will no longer be an effective strategy as it’s too blunt, and forces an exact match between the anchor text and the content – which might not actually be what the user is looking for. Moving forward, a better strategy is to use more general anchor text which is contextually relevant to the page, and let the link follow natural language processing to determine the context and relevance between the two pages.