In 2021, Google announced the development of a new AI system called Multitask Unified Model (MUM), which represents a significant advancement in natural language processing and search capabilities.
With its ability to understand context, intent, and meaning across multiple languages, MUM has the potential to greatly enhance the Google Search Experience.
However, Google MUM also poses some interesting challenges for SEO. As an AI system designed to provide direct answers to natural language questions, MUM could reduce the need for users to click through to other websites.
This could significantly impact click-through rates and website traffic, two key metrics in SEO.
What is Google MUM?
Google MUM (Multitask Unified Model) is an AI system created by Google to understand and answer difficult questions. Announced in 2021, it extends on Google’s earlier natural language systems, such as BERT, by expanding model size and training it on more data to handle multi-dimensional, open-ended queries.
Because of its huge scale, it is referred to as a “thousand-billion parameter” model. This allows it to answer questions by combining information and context from multiple papers and websites. MUM is trained in 75 different languages and for a wide range of activities, including translation, Q&A, summarization, and conversational response.
This multi-task training enables it to respond to user queries with the appropriate thinking style.
- The main goal of MUM is to deliver more relevant and comprehensive responses by recognizing nuance and context and even following queries rather than just keywords.
- It is still an experimental research system, so people cannot access it directly.
- Google intends to integrate MUM findings into consumer products to improve search, recommendations, and question-answering.
Google BERT vs. Google MUM
Google BERT and Google MUM are two important generations of AI language models developed by Google, with certain key differences.
| Google BERT | Google MUM |
| They were introduced 2018 as an “encoder” model for NLP applications such as question answering. | They were introduced in 2018 as an “encoder” model for NLP applications such as question answering. |
| Trained on over 3 billion words to fully understand linguistic contexts within text sections. | Has more than 1 trillion parameters, which is approximately 3000 times more than BERT. |
| Enables bidirectional context for words by analyzing entire sentences at once. | Equivalent to millions of knowledgeable human assistants. |
| It was announced in 2021 as the next evolution of Google’s NLP efforts. | Its multi-task architecture enables it to translate, summarise, annotate, and perform other functions within a single model. |
How will MUM affect Search Results?
MUM aims to respond to complex search questions and fit more of a customer journey onto a single SERP. Most of the advertised MUM-related features appear to be enhanced recommendation systems. Here are the features of MUM:
a. Updated Google Lens
MUM allows you to use Google Lens to capture a picture and ask a question about what you see.
b. Multilingual Google MUM
Google MUM can analyze data in 75 languages and attempt to identify the best answers to asked queries, even if the query is entered in a different language. Sometimes, the best answer to a query is not accessible in the language being searched. The MUM algorithm will search for the best result in 75 languages and return the most relevant results for the query.
c. Multitask Architecture
Unlike models trained for a single job, MUM can translate text, analyze data, generate images, offer search results, hold conversations, and do much more within a single system.
d. More Huge Images
For certain inquiries and searches, Google uses MUM to identify when visual SERPs should be triggered and which images from SEO ranking sites should be displayed.
e. In-depth Video Analysis
MUM will allow Google to analyze videos more extensively and find subjects related to different sections. As a result, Google can employ its recommendation engine to make better recommendations for what to watch, read, or consume after watching a video.
f. New Recommendation Features
Google is implementing a new SERP feature called Things to Know, an enhanced recommendation system. Its purpose is similar to the People also Ask feature: to narrow your search and move you along your buyer/learner journey.
The main difference is that the People Also Ask feature typically takes you one or two steps farther in your search, whereas the Things to Know feature strives to take you all the way.
This is not a new feature; rather, it is an evolution of an existing one, with the ideas becoming more confident and far-fetched.
g. SERP Rank Will be Less Significant
As Google MUM gains popularity, ranking in SERPs will become increasingly unimportant. In short, it doesn’t matter whether you rank first after about 10 “non-organic” search results because so many features dominate the top half of the SERP.
MUM gives the context of the query a lot more weight. This means that if two people search for the same thing, they will likely get different results. Search history, geography, buyer journey stage, and general content preferences will all impact the outcome.
How does Google MUM affect SEO?
MUM’s main objective is to answer difficult search queries while providing a positive user experience. The general availability of a super-powerful MUM upgrade will undoubtedly impact traditional SEO strategies used to rank websites.
Let’s see how MUM will affect SEO!
a. Getting Rid of Language Barriers
Language is the most significant barrier for many customers. Websites that are not bilingual appear lower in search results. With 75 languages, the MUM algorithm can overcome all of these limitations.
b. Understand Information in Many Formats
The MUM algorithm will search beyond written words and prioritize video, audio, and image content. It will assist Google in gaining deeper insights and finding more relevant content for the user’s search queries.
c. Producing Predictive Answers
This new upgrade will assist Google in guessing the complete, multi-layered search intent underlying a user’s query. The main aim is to give the best-tailored results and helpful content without asking the user a sequence of several questions. This approach ensures that the content aligns with the top Google questions that users frequently search for.
d. Integrating Multiple Sources into One
It combines multiple results to create distinct text. You will receive a unique result for your query. In brief, the outcome will be a condensed version of the most valuable and essential points.
How do you Adapt your SEO Strategy to MUM?
MUM doesn’t require a fundamental change in your SEO strategy. However, it does prioritize some of the newer SEO features, such as entities and complex content markup. The following are the strategies that will be significant in implementing MUM.
a. Do not Give Up on Traditional SEO
One of the drawbacks of complex algorithms is that they demand a significant amount of computing power.
Yes, Google can now analyze your sites in a lot more depth, but it cannot analyze every page on the web.
So, Google will likely use a simpler algorithm to pre-select pages based on relevance, quality, and authority, followed by MUM to extract specific chunks of information from these pages.
Like earlier, you must show subject relevance with entities, authority with backlinks, and quality with user experience signals. I doubt MUM will ever look at pages that do not address these topics.
b. Choose Entities over Keywords
Google used to assess page relevancy based on the number of times a specific term appears on a page. The more times the term is used, the more relevant the page is.
Establishing entities extends beyond keywords. It combines employing the correct keywords (in moderation), creating links (not simply backlinks), and applying suitable markup.
c. Emphasis on Customer Journeys
Try following Google’s approach in your content strategy. If you’re writing content for a product, you want to cover every customer journey phase.
d. Join Google Merchant Centre
Google Merchant Centre is a program that allows you to check your products and display them on various Google services. In short, if Google wants to make a product recommendation, it will select from a pool of products validated by the Merchant Centre.
Taking advantage of this program is especially necessary, given Google’s shift towards a clearer commercial agenda. By reducing buyer journeys with multimodal SERPs, Google now has an excuse to provide product recommendations for almost any sort of search, even if the searcher’s purpose is not immediately transactional.
e. Create Multi-Modal Content
MUM analyses various content types, including text, photos, and videos. To optimize for MUM, create content that uses several modalities. Combine useful text with relevant images and videos to create a more engaging user experience.
f. Separate Your Information into Snippable Pieces
A clean content structure will allow Google to extract paragraphs of text or video segments from your content. To establish a clear content structure, follow these basic concepts.
- Your content should be divided into chapters and sub-chapters.
- The chapter names should sound like search queries.
- Proper markup should be used to highlight chapters.
- Chapters should begin with concise paragraphs that fully answer the query.
Will MUM be different from other Google AI updates?
Yes, the implementation of MUM has the potential to be far more transformative for Google Search and its products than previous AI developments. It will transform the old method of processing information and determining what is best for your needs.
Users will soon be able to virtualize related topics to their primary inquiry. Finding quality stuff in one place will reduce frustration and web consumption time. That is what the network behind MUM is aiming for.
Previous machine learning improvements aimed to improve the search experience, prevent errors, and detect blackhat links and plagiarised information on the web. In a few subsequent upgrades, Google enhanced the “intent” feature. It used advanced machine learning to link the search query language to the underlying NLP processors, satisfying user intent and making Google a more dependable engine.
Previous AI upgrades, like neural matching, Hummingbird, RankBrain, and BERT, were centered on technical SEO and structured data alignment. They made a place for organic content and expert-written stuff. However, with generative AI, the focus turns to what is best for the user to see, whether organic or sponsored.
Google hopes to achieve the unthinkable by transforming SERP into a distributed social and community network. With this in-depth SEO technique, consumers will be exposed to the most recent trends and news in the industry they are searching for.
Google will not only reduce research efforts but will also supply a lot of knowledge using AI.
Conclusion
MUM marks an important milestone in AI-driven search as Google develops language models with extraordinary scalability and multitasking capabilities. MUM, trained on over a trillion words in 75 languages and many modalities, intends to give more relevant, personalized, and conversational search experiences.
Google’s MUM seeks to learn more about us than any other search engine to understand better what you may be looking for. With the arrival of Google MUM, you will be saying goodbye to BERT.
FAQs
1. How does MUM differ from previous Google AI updates?
MUM is Google’s most scalable, multi-tasking AI model to date. It can interpret using a massive linguistic and multimedia knowledge graph of over 75 languages and learn from the visual world.
2. What is the impact of MUM on Search Engine Optimisation?
Semantic search that understands query intent takes priority. As MUM evolves search, optimizing for accessibility, quality material, structured data, graphic components, FAQs, and user experience may become increasingly important.
3. How can content creators use MUM to improve visibility?
Content creators can benefit from MUM by creating high-quality, diverse content relevant to user intent. Incorporating relevant photos and videos, optimizing for context, and staying current on MUM developments can all help increase content visibility.
4. Is MUM only focusing on English-language content?
No, one of MUM’s strengths is its capacity to understand and process content in various languages simultaneously. It seeks to create a more inclusive and internationally relevant search experience for users from various linguistic regions.
English
French
German
Russian
Spanish