What is BERT? Understanding Google’s Algorithm Update

What is BERT? Understanding Google’s Algorithm Update

On the 25th of October 2019, Google announced BERT. This was introduced as a significant improvement in the way Google understands search queries. So why should you care? Well, Google themselves described BERT as the most significant leap forward for them in the past 5 years, and one of the biggest advancements in the search engine’s history.

So, let’s take a close look at BERT.

Mobile phone with two pencils representing the BERT Google update

What Does BERT Mean?

BERT is an acronym for Bidirectional Encoder Representations from Transformers. You can see why it is shortened to BERT.

Quite a mouthful otherwise!

What is BERT?

BERT is an open-sourced neural network-based technique for natural language processing (NLP). This looks at all the words in a sentence to understand a search query, rather than looking at the words one-by-one in order. The ‘intent‘ of a word is better understood by the search algorithm since it looks at the words both before and after it.

Google stated that by applying BERT model techniques that the algorithm will be able to better understand 10% of all English queries in the United States. The queries that are most relevant to BERT are longer form, more conversational queries where include ‘for’ and ‘to’ carry significant weight towards understanding the context behind the search.

This will be useful for voice search, where more conversational search queries are used.

This progression makes sense since it’s estimated that 50% of all searches will be voice searches by 2020. With this increase in voice search volume, Google needed a way to better understand the more conversational search queries that people use.


Google also provided a few examples.

In this example, you can see that the search term relies heavily on the word ‘to’ to derive the actual intent behind the search.

BERT in Search: Visa Example showing the importance of the word 'to'

Source: Google https://www.blog.google/products/search/search-language-understanding-bert

Before applying BERT, the search engine returned results relating to US citizens. The algorithm missed the importance of the word ‘to’. This reveals that the search query is actually about a Brazilian traveller heading to the USA.

After the update, Google is able to better recognise the intent behind the search and deliver a more appropriate search result.

Here’s another example focusing on the word ‘for’:

A screenshot of a Google search before and after the BERT update highlighting the example word 'for'

Source: Google https://www.blog.google/products/search/search-language-understanding-bert

In this example, the search engine could not distinguish the importance of the word ‘for’. After the update, it is able to identify that the search intent is to know whether you can pick up a prescription for someone, rather than just providing general information about pharmacy prescriptions.

Implications on SEO

Anything that changes the way search queries are understood will impact SEO. As mentioned before, Google stated that BERT will impact one in 10 searches. Some people will naturally think that ‘impact’ is negative, however, it can be positive as well.

We think that if you have been optimising your on-page SEO for voice, then BERT is going to result in a good outcome. Voice search requires a more conversational content structure, which should work well with the new Google update.

The ability of the search engine to understand words such as ‘for’, ‘do, and ‘to’ means that it be able to understand search queries that better match your content.

The websites that have been using outdated techniques for optimising keywords will run into trouble. These usually include keywords in strange arrangements that don’t follow natural language structure.

As always, all SEO efforts are undertaken within the rules of Google. So the only real thing to do is to keep an eye on the data and make adjustments where necessary.

What To Do If Your Traffic Has Been Affected By BERT

If you notice that your traffic has been affected since BERT was announced, then you should probably audit your content.

Does your content answer questions that people are searching for?

Is your content structured in a natural language framework? i.e. does it include conversational language that someone is likely to say?

Is your content thin or lacking substance?

Have you tried to ‘game’ the search engines by including mountains of keywords without actually providing anything of value?

Either way, don’t panic yet. It is still early days and it appears that BERT will be rolled out to the US in English first, with other countries and languages to follow. No doubt that this will have an impact on the future of SEO practices so the best thing to do is monitor the data and adjust your strategy as we get a better understanding of how BERT is now populating the search engine results page (SERP).

The announcement of BERT is a big step forward for Google.

It allows the search engine to better understand the intent behind a search so that they can deliver better search results. As always, any update to the search engines will impact SEO. You need to be creating content that follows natural language. The easiest way to do this is to focus on content that is created for humans.

After all, Google is trying to better serve searchers, so start with the searcher in mind.