Back to Blog
Tiles spell out search

Answering Every Question You Ever Had About BERT

Following the announcement and roll out of BERT, Google’s most recent massive algorithm update that has become a gamechanger in the field of search, some mystery still remains regarding what BERT will mean for marketers and SEO teams.

To help clear up any confusion, we’ve answered some of the top questions being asked about BERT to give you a better understanding of this revolutionary update.

What Is BERT and How Does It Work?

Referred to as “one of the biggest leaps forward in the history of Search,” and Google’s most significant update in 5 years, BERT is a pre-trained deep learning natural language framework that stands for:

– Bi-Directional

– Encoder

R – Representations

T – Transformers

Using Natural Language Processing (NLP), BERT helps Google search develop a deeper understanding of language context and flow and allows Google to process words in relation to all the other words in a sentence, rather than one-by-one.

Simply put, BERT helps computers to better understand language like humans.

For example, BERT takes into account prepositions like “for” or “to” that can dramatically change the meaning of any particular search query.

Previously, Google would have overlooked these words.

Therefore, BERT will work most effectively with searches that consist of full sentences and are more conversational in tone.

BERT also improves the way Google understands users’ search inquiries along with their intent, in order to provide more accurate results and featured snippets more quickly.

“These improvements are oriented around improving language understanding, particularly for more natural language/conversational queries, as BERT is able to help Search better understand the nuance and context of words in Searches and better match those queries with helpful results,” Pandu Nayak, Google Fellow and Vice President, Search, stated in a recent blog post.

“Particularly for longer, more conversational queries, or searches where prepositions like “for” and “to” matter a lot to the meaning, Search will be able to understand the context of the words in your query. You can search in a way that feels natural for you.”

Improving Search in More Languages

BERT isn’t just for English users, as it is able to improve search results across the world in different languages.

Another key characteristic of BERT is that it is able to take learnings from one language and apply them to others.

“We can take models that learn from improvements in English (a language where the vast majority of web content exists) and apply them to other languages. This helps us better return relevant results in the many languages that Search is offered in,” stated Nayak in the BERT blog article.

Backstory

In the past, NLP models have consistently struggled to differentiate words based on context.

Text analytics historically relied on shallow embedding methods where a given word could only be assigned one vector.

As an example, the vector for the word “wound” would have needed to include information about clocks as well injuries.

BERT, however, maps vectors onto words after reading the entire sentence.

How Will BERT Impact Featured Snippets?

Besides improving overall search results, BERT is also being used globally to improve featured snippets in all languages.

Google hasn’t revealed how exactly BERT improves featured snippets, so we can only assume that it’s similar to what BERT is doing overall, which is gathering more relevant results based on the context of a particular search query.

Here is an example of how BERT helps Google to display more relevant featured snippets:

 

Google featured snippet

Source: Google

Before BERT, the search query shown above query would confuse Google’s systems.

Google explains that they “placed too much importance on the word “curb” and ignored the word “no”, not understanding how critical that word was to appropriately responding to this query. So we’d return results for parking on a hill with a curb.”

How Will BERT Impact Other Google Products?

As of right now, BERT only pertains to search, however, it will inevitably have an impact on Google Assistant as well.

Results and featured snippets triggered by search queries conducted via Google Assistant may be influenced by BERT.

Google has also said that BERT isn’t currently being used for ads, but there is a possibility that it may be integrated in the future to help alleviate some of the bad close variants matching that often afflicts advertisers.

BERT Vs. RankBrain

BERT has been described as Google’s most important update since RankBrain, Google’s first artificial intelligence method for understanding queries introduced back in 2015.

RankBrain looks at both queries and the content of web pages in Google’s index to better understand the meanings behind the words being used.

It’s important to note that BERT does not replace RankBrain.

via GIPHY

BERT is simply an additional method for understanding content and queries, and RankBrain will still be used for certain queries.

In fact, multiple methods – including BERT – can be used to help better understand a single query.

What BERT Means For SEO

Unlike previous algorithm updates, BERT does not judge web pages either negatively or positively. Instead, it improves Google’s understanding of human language when producing search results.

And as a result of BERT, Google is better able to understand the meaning of content on pages it comes across along with the context behind user’s search queries.

Therefore, “There’s nothing to optimize for with BERT, nor anything for anyone to be rethinking,” says Google’s Public Search Liaison, Danny Sullivan. “The fundamentals of us seeking to reward great content remain unchanged.”

So, What Can Be Done? – Google’s Advice

Google’s advice has always been to create quality content that matches searcher intent and continues to be their main recommendation post-BERT.

However, during a recent online chat, Google Webmaster Trends Analyst, John Mueller also recommended that sites that have experienced a decline in rankings due to BERT should focus on the fundamentals of SEO.

“My recommendation here would be to not focus on BERT, not focus on purely technical aspects…it’s something where I would really focus on the site overall and kind of improving things overall,” said Mueller.

How to Put Google’s Advice into Practice

Despite the fact that Google’s official position is to not to optimize for BERT, there are ways to boost your SEO in light of this update, as per Mueller’s recommendation.

Optimize for Humans Instead of BERT

It’s been said before, but it deserves repeating: Write for humans, not machines.

If you’re already putting out high quality content written for humans, congratulations. You’re already one step ahead. But if your strategy still focusses on keyword stuffing, take note.

Don’t Forget to E-A-T

SEO best practices haven’t changed with introduction of BERT, so it’s important not to forget the fundamentals such as E-A-T.

E-A-T stands for Expertise, Authoritativeness, and Trustworthiness and refers to a broad core Google algorithm update that reinforced the importance of creating content from trustworthy resources that are experts their field.

So, along with writing for humans, continue to watch how you E-A-T.

Look at How Your Site Captures Informational Searches

One of the most noticeable side effects of BERT that marketers have noticed is a drop in organic search traffic, suggesting that before the update, pages were ranking in searches that weren’t applicable to their niche. So how do you combat this?

Because BERT is designed to provide more relevant search results to users, you’ll need to go back and:

  • Identify what keywords or pages are no longer performing as well as they were before BERT.
  • Search those keywords to determine what other content is now appearing either alongside or instead of your content and identify any changes.
  • Make adjustments to your content to better match the ranking content or start producing more content to reflect these changes.

Optimize for Featured Snippets

While you cannot create your own featured snippets, there are ways you can help increase your chances of appearing as one.

Here are some tips for optimizing your content to appear as featured snippets:

  • Organize your content in lists
  • Answer questions in 40 to 50 words or less
  • Use tables to display data
  • Produce more video content

So, there you have it. While it’s true that BERT is a pretty significant update that will have a major impact on both users and marketers, search engines still have a long way to go, and BERT is just a small part of a never-ending puzzle.

But, by writing quality content created with humans in mind and continuing to follow SEO best practices, you’ll remain in good shape as newer advancements inevitably arrive.