Bidirectional Encoder Representations From Transformers. Imagine having to use that in a sentence! Let’s keep it to BERT for short.
BERT is the latest official Google update to their search engine algorithm, rolled out at the end of October 2019 and impacting English-language search queries only for the time being.
In Google’s own words:
“[BERT] represent[s] the biggest leap forward in the past five years, and one of the biggest leaps forward in the history of Search.”
So, What Is Bert Exactly?
BERT is Google’s attempt to understand the context of any given search query. Rather than looking at individual words in a query, BERT attempts to process the search term as a whole.
Google’s first foray into contextual understanding was with their Hummingbird update in 2013. The Hummingbird update looked at the search query as a whole, rather than just as individual words. The update was Google’s attempt at interpreting the intent behind what internet users were looking for in their searches.
Users nowadays are using more and more conversational terms for their searches as devices have become much more than just a smartphone or other web-enabled device. With the rise in voice search, contextual search delivers the most relevant results.
How Does Bert Work?
In a word, Transformers. Not cars that become robots, but analytical models that attempt to process the words in a sentence in relation to all the other words in that sentence, rather than stacking them one-by-one in the order they appear in the sentence.
BERT works by looking at the sequence words appear in a sentence as it tries to understand the context of the search. For longer, more conversational searches and searches where propositions like “for” and “if” have an impact on the context, BERT will help to search more naturally.
Here’s an example straight from Google. The search term “2019 brazil traveler to usa need a visa” is a search for information concerning a Brazilian looking into visa requirements for travel to the USA.
Before BERT, the Google search algorithm would have returned results related to U.S. citizens wanting to travel to Brazil. With BERT, Google’s algorithm understands that “to” plays a significant part in the context of the search.
Here’s another example from Google themselves. Compare the search results for the term “can you get medicine for someone pharmacy.”
BERT now understands that “for someone” is an essential part of the search. The searcher is looking for information about filing a prescription at a pharmacy for someone else, as opposed to just information about filling prescriptions in general.
For marketers, BERT will probably have the most significant impact on featured snippets. These are the results Google deems to offer the best answer for a specific question and make for some incredibly valuable SEO real estate.
With BERT, these will carry the best answers to the context of a given search rather than the word stacking in the search term itself, with the update being rolled out globally for all languages for featured snippets.
As the update beings to impact more and more search terms, users will navigate more towards naturally phrased terms, rather than the keyword-Esque terms they use now. We won’t need to modify the way we use words in a search term to get Google to understand what we’re looking for as BERT will do that for us.
Google estimates that the first update will impact something like one in ten searches in English, with the update being applied to more languages over time as the data comes in, and BERT evolves.
So How Do I Optimize For BERT?
If you’re publishing relevant, engaging, and informative content, you’re good to go. You don’t need to do anything else. While BERT wasn’t created to penalize sites, it is conceivable that you may notice a drop in traffic. This drop isn’t a penalty; there isn’t anything to recover from doing. In most likelihood, your site was getting clicks for searches that shouldn’t have been getting to you in the first place and which would have been pretty low quality in any case, with subsequent bounce rate increases. What traffic you do get will be more relevant and more likely to engage with whatever you have to offer.
While BERT isn’t something you can optimize for, it’s not something you can afford to ignore, either.
There used to be a medium for content, and that was the line between what Google would interpret as being relevant and what your visitors would find relevant and useful. With BERT, that line is fading. What Google can understand and what visitors are looking for, now become the same.
If you create content that answers specific queries, in the context these queries addressed, then BERT can only be a benefit to you. If you have seen a drop in traffic, chances are this hasn’t made much difference to your conversions. Drill down into the landing pages that saw the drop-off, and you probably won’t see any decline in conversion figures.
The upshot of this is that what traffic you do get will be much more prone to convert. After all, you’ve answered the exact search query. Why wouldn’t you get the conversion?