MUM

MUM stands for Multitask Unified Model and refers to the AI feature in Google Search, specifically aimed at improving search capabilities across multi-language editorial and multimedia content to provide people with answers to the more elaborate questions.

What is MUM?

Introduced in 2021, MUM is the updated Google Search algorithm, which is to replace the previous iteration, known as BERT (Bidirectional Encoder Representations from Transformers). 

From the technical standpoint, MUM uses the so-called T5 text-to-text framework and implies several essential upgrades, regarding the deeper content analysis across multiple languages, and ability to complete search results with various types of multimedia content (images, video content, podcasts, etc.). 

Key MUM Features

One of the key MUM features is the so-called topic zooming, i.e. users’ ability to switch from one subtopic to another more smoothly, i.e. by shifting between the “Refined” and “Broadened” search results, if needed.

The search results will also include the new “Things to Know” section, where the engine will be offering a selection of answers to some of the most probable follow-up questions, supposedly making the related search faster and easier. 

In addition, a list of search results will also feature a set of the so-to-speak visual aids, represented by browsable image results along with text links.

More importantly, the MUM integration with Google Lens will also enable asking questions while snapping a photo, and getting shoppable items in the search results, accordingly.

Questions Yet Unanswered

While Google’s MUM isn’t likely to roll out in full until 2022, a lot of questions regarding its potential impact on websites’ ranking and their display in search results, remain unanswered. 

What is already clear, however, is the greater importance for publishers to ensure their editorial content actually provides valuable information and insights to users, exploring a particular topic, while also taking leverage from other types of content (image, audio and video) to answer their audiences’ potential questions.


Back to Glossary