No, Google’s BERT update has nothing to do with the Sesame Street character. BERT stands for Bidirectional Encoder Representations from Transformers, which is a mouthful (and the name does describe what it does). It was Google’s fall 2019 update, and it gave the search engine a far better way to read what people type into the search box. Instead of matching a query one word at a time, Google could now read the whole query in context and return results that fit what the person meant. The name is hard to remember, but the update itself is not hard to follow. Below we explain what BERT is, what it changed and how to write your SEO content so it works in your favor.
Launch Date: October 21, 2019
BERT started rolling out the week of October 21 for English-language queries. It changed how Google reads a search: as a sentence with meaning, not a string of keywords.
What BERT changed in Google search
Google said BERT would affect about 1 in 10 English-language searches in the US. That sounds small, but you want every search that could reach you to count. The good news is that BERT does not penalize sites or pages. Even so, there are two common reasons a page might rank below a competitor’s after BERT:
The wrong keywords
Keywords still count. If you target only a narrow list of terms, BERT will tend to favor pages that use the words and phrases people actually type when they search.
Content that reads unnaturally
BERT was one of the most advanced changes Google had made to its algorithm, and it reads context and intent far better than earlier systems did. As a result, people can type a query the way they would say it out loud, and Google can work out what they want and find a page that answers it. If your content is stiff, awkward or packed with jargon, it is harder for Google to match it to those natural-language searches. Chances are your rankings will show it.
How BERT reads a search query
BERT was a real step forward in how Google processes language. It is a technique built on artificial neural networks, and it helps computers understand human language closer to the way people do. Put simply, it lets Google figure out the context of a query, interpret what the searcher meant and return better results as a result of that.
Before BERT, Google tended to read a query word by word. BERT, a natural language processing model, reads the full query at once and looks at each word in relation to the words before and after it (that is the “bidirectional” part of the name). It treats the text the way a person would: as one thought. Small words such as “to” and “for” can change what a query means, and BERT picks up on them. Context decides the result now. That is good news for searchers, because BERT understands intent better than any update Google had released before it.
How to adjust your SEO after BERT
As we said above, BERT does not penalize anyone. Its job is to understand what a searcher wants, with more nuance than before, not to find “bad” sites and push them down. Still, there are a few small changes to your site and content that give you the best chance of ranking well under BERT.
Write long-form content
When you create content for SEO, the aim is to answer the questions your audience asks and give them the best answer to their problem. One way to become the best resource for them is to publish in-depth, long-form content that is worth reading. A longer page also gives you room to cover a topic in the words people use, so your phrasing has a better chance of matching the questions and queries they type into Google.
Find out why your rankings dropped
If you manage your own SEO, track your rankings in Google Search Console. Compare them over time to see which keywords or queries moved and when they moved.
Search Console is a useful habit, but BERT may not be the reason your rankings changed. Google changes its core algorithm many times a day as it works to give searchers a better experience. Some movement in your rankings is just part of the normal ebb and flow of a search engine the size of Google, which never stops changing.
Update your content
If your rankings drop, revise the page. Answer the questions in a different way, try new terms, come at the query from another angle or add other content types, for example images or video.
What BERT means for your content
BERT made Google smarter and better at reading language the way people use it. People search for more information every year, and there is more of it to search. A system that reads queries in context gives searchers results that fit what they asked better than anything Google had before 2019.