Google released a new artificial intelligence (AI)-based update to its ranking algorithm in the summer of 2021. Known as the Multitask Uniform Model (MUM) update, it’s the successor to the Bidirectional Encoder Representations (BERT) update. The MUM update will forever change the way in which Google processes search queries and, thus, ranks websites. To attract search traffic from the most popular search engine, you must learn how MUM will affect your website.
What Is MUM?
MUM is an AI system that Google recently incorporated into its ranking algorithm. It uses an intuitive, multimodal approach to processing search queries and ranking websites.
MUM is essentially a natural language processing (NLP) system like BERT. It leverages AI to understand the intent of search queries rather than simply the definition of the words in search queries.
Google no longer looks exclusively at the definition of individual words in search queries. Instead, it focuses on intent. Google seeks to understand what users are trying to find by entering a given search query. With MUM, Google can interpret the intent of otherwise complex search queries.
MUM uses AI to understand search intent so that Google can return high-quality and relevant results.
Not all search queries consist of one or two words. Many of them consist of a half-dozen or more words. To process complex search queries such as these, Google uses MUM. MUM is baked into Google’s ranking algorithm. When a user enters a search query, MUM will evaluate the context and relationship of the words to understand what the user is trying to find.
MUM vs. BERT: What’s the Difference?
MUM and BERT are two of the most significant Google updates in recent years. Google released BERT in October 2019, whereas MUM came several years later in May 2021. Aside from being newer, though, there are other ways that MUM differs from BERT.
As an NLP system, MUM is more powerful than BERT. According to Google, MUM is 1,000 times more powerful than its predecessor. It can interpret search intent — specifically the intent of complex search queries — with a higher level of accuracy than BERT.
One of the reasons MUM is so much more powerful than BERT is its ability to understand multiple languages. MUM has been trained to understand 75 languages. In other words, it pools information from sources in all of these different languages to process search queries.
MUM doesn’t just use sources in the language of the search query; it uses sources in 75 different languages. BERT cannot understand multiple languages, resulting in weaker and less-accurate query processing.
While BERT is limited to text, MUM supports other formats. Google revealed that MUM could be applied to images and videos. MUM is multimodal, meaning it can understand information in multiple formats. Google is planning to integrate MUM into Google Lens, for instance.
Available at lens.google.com, Google Lens allows users to perform searches by uploading or specifying an image. Google Lens will retrieve information about the uploaded or specified image from the internet. If a user wants to find a store that sells a particular product, the user can upload an image of the product to Google Lens. MUM is expected to power Google Lens in the near future.
How MUM Will Affect Your Website
You might be wondering how MUM will affect your website. For starters, MUM may result in your website ranking for longer search queries than it did in the past. With MUM, Google can process complex search queries. Longer search queries, of course, are more complex than shorter search queries.
You don’t have to optimize your website only for short and simple keywords. MUM can use sources in different languages and formats to process complex search queries. As a result, you can optimize it for longer keywords. When optimized for a relevant complex keyword, your website will likely rank for it.
MUM places a greater emphasis on visual forms of content. Unlike BERT, it can understand images and videos. Therefore, your website may rank for more forms of visual content now that MUM powers Google’s ranking algorithm. Your website’s images may rank in Google’s “Images” tab, and your site’s videos may rank in the search engine’s “Videos” tab.
Of course, visual content requires optimization. Uploading a high-quality image or video to your website isn’t enough. For MUM to understand it, you must add context. You can create a title and alt text description tag for images. Titles are displayed when visitors hover their mouse over the image, whereas text-to-speech software uses alt text description tags. Regardless, you can use them to add context so that MUM can understand the images on your website.
There are tags with which you can optimize videos, such as title, duration, description, and URL. Optimizing a video with these tags will help MUM understand it. Just remember to create relevant tags that describe the videos. Irrelevant tags will only harm a video’s ability to rank.
Google uses MUM to process local search queries. Unless a local search query contains the name of a place, Google will assume the user is looking for information related to a nearby business or landmark. But tourists may perform local searches in their native language rather than in the country they are visiting.
MUM is multilingual. Suppose a tourist searches for a local business or local place in their native language. In that case, MUM will use sources in a variety of languages to return relevant local listings. It’s able to understand 75 languages. MUM will scour the internet while looking for information about the search query. Even if a piece of information is in a different language, MUM will use it.
Prior to MUM, Google relied on BERT to process complex search queries. BERT was a major stepping stone in the evolution of Google’s ranking algorithm. It allowed Google to interpret search intent. MUM uses a similar approach to processing search queries but is 1,000 times more powerful.