Targeting specific keywords as an SEO strategy is never going to serve as a long-term answer for ranking in search. Instead, your goals in content creation and SEO should focus on the intent(s) or types of searches instead.

In late October of 2019, Google announced the rollout of its newest algorithm update, which further proves why this strategy is the future of search: BERT.

What is BERT?

More like, who is BERT? BERT’s name is an acronym that stands for Bi-directional Encoder Representation for Transformers, and it is artificial intelligence (AI) created by Google. BERT’s job is to greatly improve search results by helping Google understand the context of search queries. As of its launch, BERT only speaks English and is only available in the United States. We do expect BERT to grow beyond its initial capabilities.

The Need for BERT

Google used to take search queries like “Double Jeopardy” and give you results that included “double”, “jeopardy”, and “jeopardy double”. Since there are many different meanings behind “double jeopardy,” Google couldn’t do much more than hope that a searcher would find a relevant result.

The term, “Double Jeopardy” is the name of a law, but is the searcher looking for information about the law? Has the law been applied to a trial that happens to be trending in the news? Is the searcher looking to stream the movie starring Ashley Judd and Tommy Lee Jones? Perhaps the search is regarding the very popular game show “Jeopardy.” Searches that contain such a diverse mixture of possible results are very difficult for any computer program to decipher, which makes the computer’s natural language processing (NLP) extremely important.

Example of BERT in Action

Search query: How to make pancakes from scratch

Starting with the subject of “pancakes,” the BERT algorithm might break down the user’s query like this:

How to make pancakes from scratch

BERT: Search Google database/index and find all pages related to pancakes

BERT: Do humans use any other words that have the same meaning as pancakes?

BERT: Yes. Pancakes also Include crepes, hotcakes, flapjacks, etc.

How to make pancakes (crepes, hotcakes, flapjacks) from scratch

BERT: In context, “How to make” indicates that the searcher expects instructions and recipes. So, eliminate web pages related to restaurants, where to buy, reviews, movies that include pancakes, etc.

The new query looks more like:   

How to make (instructions, recipes, ingredients) pancakes (crepes, hotcakes, flapjacks) from scratch

BERT: The query string from scratch indicates to exclude commonly known mixes and include raw cooking materials.

In this example, BERT was able to breakdown human language to better understand the complete context of the search by including relevant content and excluding irrelevant information.  

Next, Google applies its RankBrain and PageRank formulas to determine the best possible web pages to satisfy the end-user. 

How to Optimize for BERT

You can’t exactly optimize for BERT. BERT is a machine-learning algorithm that learns more things every day. Your job as an advertiser is to optimize your content for your end-user. Know exactly what your audience is searching for and provide the most comprehensive details possible, plus help them take the next steps on your website. Through detailed content, you can help teach BERT how your users speak and use search.

If you’re interested in learning more about BERT, content marketing, or SEO, contact our SEO experts today for an SEO consultation.