Google MUM: toward a better understanding of search intent

On May 18, 2021, Google’s vice president of search, Pandu Nayak, announced the arrival of Google MUM. This new algorithm, 1,000 times more powerful than BERT, should allow the search engine to better understand user intent and to make sense of complex queries.

Here’s a closer look at what promises to be a major revolution for SEO.

What is Google MUM?

Multitask Unified Model: MUM is THE new version of Google’s algorithm, based on Artificial Intelligence and designed to replace the current algorithm, BERT, in the near future. The latter, an acronym for Bidirectional Encoder Representations from Transformers, is a computer language model capable of grasping the meaning of words based on their position within a sentence. The BERT algorithm therefore already understands the contextual relationship between words and is able to give meaning to sentences.

Announced as 1,000 times more powerful than BERT, MUM will be capable of understanding and interpreting the meaning of an image, and even the meaning of a video or an audio file. The new algorithm will thus make it possible to provide search results that are even more relevant for internet users.

One of MUM’s great strengths is above all its ability to handle and respond to complex queries in a precise way. Indeed, the development of the algorithm follows several observations made by the American company’s leaders:

Internet users perform an average of 8 searches to find the information they’re looking for.
Queries are getting longer and longer: 50% of searches contain 4 words or more.
Voice search, particularly via smart personal assistants, is intensifying and requires an immediate answer.
15% of daily queries have never been typed into Google before.

The search engine’s goal is therefore to go even further in interpreting users’ search intent and thus to help them find the answer to their question more quickly. Do you want to be sure you’re producing content that’s useful for internet users? Download our template for free to interpret THE right search intent behind every query and drive as much traffic as possible to your content.

How does Google MUM work?

Illustration: google mum 1

Source: https://diginoam.com/

Google MUM’s great strength is therefore its ability to respond to complex queries. To achieve this, the algorithm relies on several unprecedented skills that give it a better understanding of information.

The American company has so far released only a few details about exactly how the algorithm works. We only know that AI will play an extremely important role in how it operates.

Nevertheless, the company’s leaders were keen to demonstrate the power of Google MUM and its ability to understand and interpret with the following query:

» I’ve hiked Mount Adams and now I want to hike Mount Fuji next fall, what should I do differently to prepare? «

Below, you’ll find the video demonstration.

This query is complex because it requires the algorithm to:

understand the meaning of the question: this is a search about preparing for a hike,
compile the various data about the two mountains,
retain and compare the most relevant information (e.g., altitude, weather conditions, preparation and equipment for hiking Mount Adams and Mount Fuji, etc.)
take into account the data on the season,
offer an appropriate answer.

To handle this type of query, the company developed an algorithm with 4 major characteristics:

Multitask: it can handle several tasks at the same time (analyzing a video, answering, finding content in several languages, etc.)
Ability to understand and generate language
Multilingual, it works in 75 languages
Multimodal, it no longer analyzes only text content but also videos, images, and audio content.

It’s thanks to these 4 skills that Google MUM can analyze and offer the following answer:

“Mount Fuji is roughly the same altitude as Mount Adams, but fall is a rainy season on Mount Fuji. You’ll probably need a waterproof jacket.”

To complement its answer, the search engine also offers text, visual, and audio content to help prepare for a hike on Mount Fuji.

More than a simple results page, Google MUM therefore provides a direct answer whenever possible and also offers a wide variety of content and data from multiple sources: articles, images, videos, audio. It can also suggest content in several languages if it perfectly matches the user’s query.

Illustration: gif mum

Illustration: scren mummy

Source: https://blog.google/products/search/introducing-mum/

It’s no longer just about surfacing the best pages, but about understanding the entire context of the search across all media and all languages, and finding the best answer.

Google MUM: what will it change for SEO and how to prepare for it?

Still in the testing phase, Google has not yet announced any release date for its new algorithm. Nevertheless, Google MUM is in line with the various developments of the search engine following the BERT update in 2019 and the launch of Google SMITH in 2020.

For users, Google MUM should make it possible to:

Find an answer more quickly, including on a long query,
Improve voice search,
Obtain content and resources in several languages,
Get more relevant results in a variety of formats.

Google MUM: our SEO recommendations

While we don’t yet know the real impact this new algorithm could have on organic search, it still seems important to prepare for it. And there’s no need to panic. This preparation is a continuation of the best practices already recommended for BERT for many years, namely:

Focus on highly qualitative, user-centric content that’s as comprehensive as possible,
Improve E.A.T.,
Concentrate more on search intent,
Create semantic universes / topic clusters: a topic covered through different pieces of content in order to provide as many answers as possible on specific subjects,
Optimize photos and videos.

While waiting for Google to publish its guidelines and advice for users, you can check out our various webinars and white papers on SEO best practices: