Google announced at Google I/O earlier this year that it is investigating a new technology called MUM (Multitask Unified Model) internally to help its ranking systems better understand language.
MUM, dubbed “a new AI milestone for understanding information,” is intended to make it easier for Google to answer complex search queries.
MUM, according to Google, is 1,000 times more powerful than its NLP transfer learning predecessor, BERT.
It employs the Text-To-Text Transfer Transformer model (T5) to reframe NLP tasks into a unified text-to-text format and develop a more comprehensive understanding of knowledge and information.
MUM can be used for document summarization, question answering, and classification tasks such as sentiment analysis, according to Google.
Clearly, MUM is a major priority within the Googleplex – and something that is important to the search team should be on the radar of the SEO industry as well.
Is it, however, a ranking factor in Google’s search algorithms?
MUM As A Ranking Factor, According to the Claim
Many people who read about MUM when it was first announced naturally wondered how it would affect search rankings (especially their own).
Google makes thousands of changes to its ranking algorithms each year, and while the vast majority of them go unnoticed, some have a significant impact.
One such example is BERT.
Google called it the most important update in five years when it was released globally in 2019.
And, indeed, BERT had an impact on about 10% of search queries.
Another example of an algorithmic update that had a significant impact on the SERPs is RankBrain, which was released in the spring of 2015.
Now that Google is discussing MUM, it’s clear that SEO professionals and the clients they serve should pay attention.
Roger Montti recently published an article about a patent that he believes could provide more insight into MUM’s inner workings.
If you want to see what’s going on under the hood, this is an interesting read.
For the time being, let us just consider whether MUM is a ranking factor.
MUM as a Ranking Factor: Evidence
When RankBrain was released, it wasn’t announced for another six months. Most updates aren’t even announced or confirmed.
Google, on the other hand, has gotten better at sharing important updates ahead of time.
For example, BERT was first announced in November 2018, was made available for English-language queries in October 2019, and was made available globally later that year, in December.
We had even more time to prepare for the Page Experience signal and Core Web Vitals, which were announced more than a year in advance of their eventual rollout in June 2021.
Google has already stated that MUM is on its way, and it will be a big deal.
But, could MUM be to blame for the drop in rankings that many sites experienced in the spring and summer of 2021?
MUM’s Evidence Against Its Use As A Ranking Factor
Pandu Nayak, Google Fellow and Vice President of Search, made it clear in his May 2021 introduction to MUM that technology isn’t a factor. In any case, not yet:
“Today’s search engines aren’t quite sophisticated enough to answer the way an expert would. But with a new technology called Multitask Unified Model, or MUM, we’re getting closer to helping you with these types of complex needs. So in the future, you’ll need fewer searches to get things done.”
The timetable for when MUM-powered features and updates would be available was given as “in the coming months and years.”
When asked if the industry would be notified when MUM goes live in search, Google Search Liaison Danny Sullivan replied in the affirmative.
Nayak recently explained how Google uses AI in Search and wrote,
“While we’re still in the early days of tapping into MUM’s potential, we’ve already used it to improve searches for COVID-19 vaccine information, and we’ll offer more intuitive ways to search using a combination of both text and images in Google Lens in the coming months.
These are very specialized applications — so MUM is not currently used to help rank and improve the quality of search results like RankBrain, neural matching and BERT systems do.”
He also stated that any future MUM applications will be subjected to a rigorous evaluation process, with special emphasis on the responsible use of AI.
MUM as a Search Ranking Factor: Is Google’s MUM a Search Ranking Factor?
Bottom line: MUM is not used as a search ranking signal by Google. It’s a language AI model based on Transformer, Google’s open-source neural network architecture.
Google will train MUM on large datasets, as it did BERT, and then fine-tune it for specific applications on smaller datasets. This is what MUM’s use for improving vaccine search results is testing.
Google has mentioned specific uses for it in the (near) future, including:
- Providing insights based on its extensive knowledge of the world.
- Bringing up useful subtopics for further investigation.
- Language barriers are being broken down by transferring knowledge across languages.
- Understanding information in multiple formats at the same time, such as webpages, images, and more.
How are you going to optimize for MUM?
That is yet to be determined.
What is certain is that Google search’s intelligence is increasing by leaps and bounds.
Attempts at trickery and manipulation will become less and less effective as Google’s search algorithms become more sophisticated and capable of determining the intent and nuance of language (and likely easier to detect).
With a 1000x more powerful NLP technology than RankBrain on the horizon, optimizing for human experience is more important than ever.
If you want to stay ahead of MUM, think about what the content you’re creating means for the people whose needs it’s supposed to meet.
Machines are getting closer and closer to fully experiencing that content as your intended reader/viewer does.
Read The Google MUM Algorithm is capable of more than just ranking websites.
