From BERT to MUM: How Your SEO Strategy Needs to Change

author Team Trivera by Team Trivera on Jun 30, 2021

Google makes search engine updates and rolls out new features every year. Some algorithm changes are minimal, others extremely disruptive. Most are not explained. 

Because more than 90 percent of all searches are conducted on Google or YouTube, these changes create both frustration and opportunities for digital marketers. In 2019, Google rolled out BERT, one of its most significant changes ever to the way Google processes and understands internet searches. At its May developer conference, the company announced its next steps in the use of artificial intelligence for search - MUM.

These updates are game changers. And huge opportunities. Businesses interested in educating their customers and providing helpful content can once again engage in good, solid marketing rather than relying on keyword stuffing or other SEO tricks.

BERT explained

BERT stands for Bidirectional Encoder Representations from Transformers. This gobbledygook basically means Google is using artificial intelligence to understand how the words in a query string relate to each other rather than looking at the meaning of each word individually. At the time of its rollout, one of Google's examples of BERT's impact explained how a pre-BERT search for "how to park on a hill with no curb" would return a search about parking on a hill with a curb. The search engine didn't understand how the word "no" modified the meaning of the words around it. With BERT, the search engine result was much more accurate, returning a result that matched the intent of the query. 

BERT was a game-changer because it is impossible to optimize for it. There are no tricks to beat BERT. Marketers simply need to write content to answer the questions or meet the needs of their audience. 

The next step: MUM

Naturally, Google didn't stop with BERT. An October 2019 blog post from Google defined BERT as, "a neural network-based technique for natural language processing (NLP) pre-training." Note the use of the word "pre-training." BERT was only the beginning. Google's Multitask Unified Model (MUM) is coming, and it is evolving search far beyond simple contextual queries.

Google says MUM is a thousand times more powerful than BERT. It expands on BERT's ability to understand language and context. MUM also brings multitasking to search. A search using MUM can read and understand text, incorporate information from video and images, and combine data from over 75 languages to provide answers with results so comprehensive, you'd have to conduct multiple searches today to get the same result. 

According to Search Engine Journal, here's an example of how MUM will work: To provide a comprehensive answer to a query such as, "What are the differences between bodies of water like lakes, rivers, and oceans?", Google needs to explain what each body of water is. It then needs to compare and contrast each body of water to help the searcher understand the differences. MUM can do that. Plus, MUM understands the question even though it is phrased the way one person would ask another person, not the way we would traditionally ask a computer. In other words, Natural Language Processing, or NLP, is at work. NLP is useful specifically because many internet users conduct search using their voices, rather than a keyboard. We don't speak the way we typically string keyword-rich searches together. MUM understands that.

How to optimize for MUM

Like BERT, optimizing for MUM isn't about SEO trickery. We believe the best course of action to prepare for MUM is to create helpful content that meets your target audience's needs and answers questions that help them live better lives. And by "create content" we don't just mean a one and done blog. Adopt a content cluster strategy, creating multiple types of content that approach issues from different angles and link this content together on your website. 

Do keywords matter anymore?

Yes, keywords still matter. After all, searches are made up of words. It is important to remember that the words your prospective customer might use when searching often aren't the industry-specific, maybe even jargon terms you use. You also need to understand the basic needs of your customers and what is happening in their lives that causes them to need your product or service. If that sounds like good old-fashioned marketing, well, that's because it is. Although all the acronyms might sound like alphabet soup, BERT, MUM, NLP, and AI are here to make the internet more helpful for us all. 

Photo Credit: Adobe Stock 


Share this article