Google’s Update To The Multitask Unified Model

Stephen McCance October 9, 2021

In the last week, one of the leading announcements in SEO was the changes that have been made to the MUM (Multitask Unified Model) technology. This update was first introduced in June 2021, with the aim of improving Google’s ability to answer user’s complex search queries. 

This was a significant improvement when it came to user experience, as Google had the ability to understand an entire search query rather than simply isolating particular elements. The user intent behind the search query could be understood much better following the MUM update, which is said to be 1000 times more accurate than its predecessor, the BERT update, in 2019. 

Further improvements have been made to the Multitask Unified Model this week, providing users with an opportunity to search in new ways, which we are going to introduce to you. Regular algorithm updates and learning about new visibility opportunities is one of the main parts that we love about our job, so our team continually adapts and it is time for you to as well! 

An Introduction To The Updated Multitask Unified Model

There are four main features to the update of the Multitask Unified Model, being the combination of Google Lens and MUM technology, an improved understanding of video content, the new “Things To Know” feature and refined and broadened searches. 

Google Lens and MUM Technology

Google Lens has been combined with the MUM technology in the newest update, meaning that a user can now upload a photo along with a query, to receive relevant search results. Google Lens is an image recognition technology that allows a user to upload a photo and the contents of the image can be understood and identified. So, the combination of these technologies has the potential to completely change user experience. 

An example given by Google is if a user uploads a photo of a t-shirt with the query “find me socks with the same pattern as this t-shirt”, Google will be able to understand the image along with the query, to provide a relevant and accurate result. This is a significant part of the new generation of MUM technology, allowing users to find what they’re looking for when an accurate description with words is difficult to find.  

Improved Understanding of Video Content

Another component of the updated MUM technology is the ability of Google to identify relevant topics in videos that may not be explicitly mentioned, primarily on YouTube to begin with. The expansion to more complex modalities like video and audio is an exciting development that has the ability to allow for a more in-depth understanding of topics. This is done by identifying key moments in a video and providing more relevant topics based on a comprehensive understanding of the intent behind the video content. This update will give users the opportunity to learn more in an accurate way when they are watching video content. 

The “Things To Know” Feature

The new “Things To Know” feature provides an advanced method of allowing users to explore relevant sub-topics to their original query. Google now has a greater understanding of the different dimensions behind a search query and provides relevant information in relation to that, helping users to access an in-depth understanding of the topic they have searched for, if they wish to access more information. 

This new feature is created using technology that understands how people typically explore topics and what their subsequent searches are likely to be. Based on this, a “Things To Know” section will appear on the SERP, similar to “People Also Ask”. Rather than related questions, there will be a small drop down list with various different sections that mimic subsequent searches. Google’s example is if someone searches “acrylic painting”, the “Things To Know” section would include a snippet of an article with a step by step to starting acrylic painting, different styles of acrylic painting, how you can do acrylic painting at home, tips for acrylic painting and how to clean acrylic paint. 

This feature will give users the opportunity to find their desired answer in far fewer queries, whilst also developing their understanding of the topic should they require it.

Refined and Broadened Searches

The final change to the MUM update is that searches will be refined and broadened, acting like search suggestions in SERPs, to either allow users to gain a deeper insight into a topic or to gain more context by introducing relevant, wider topics. This ability to either narrow or widen results based on the primary search query is interesting, helping users to follow the path they were hoping to based on a brief initial query. 

Continuing the “acrylic painting” example, there would be two sections on the SERP, both a “Refine This Search” and “Broaden This Search”. The refine section would feature queries like acrylic painting ideas, acrylic painting techniques and acrylic painting online courses. The broader section would then explore topics like styles of painting and famous painters. Without the need for additional queries, the user can explore the topic either more broadly or in more detail. 

Final Thoughts

These new updates to the Multitask Unified Model could transform search for users. Although these new components are not completely disparate from what users are accustomed to, this is an exciting development that should allow users to have their complex search queries understood and answered in a much shorter time, as well as giving them the information required to explore a topic. It will be interesting to see how this theory translates into practice moving forward!

Screenshot from Google’s twitter for Things To Know section