04/10/21 Beu Smith

Google's Search On Round Up: Google's MUM & Search Enhancements

Google held its ‘Search On’ event on Wednesday where it announced some key changes to how search will be expanding. From initial interpretations these updates won't radically change the way search works but it does expand on Google’s AI and understanding the context of what a user is searching, plus refinement and expansion of your search based on your query.

Here is an executive summary of the changes:

  • Google’s MUM (Multitask Unified Model)
  • Google expands SERP features

Google’s MUM Broken Down (Multitask Unified Model)

  • Built on Transformer Architecture (Similar to BERT) 
  • 1,000x more powerful than BERT
  • AI able to multitask in over 75 languages 
  • Capable of multitasking in multiple search modalities (text, images, video etc.)
  • Deep knowledge of the world and word context 

Google’s expanded SERP features

  • Larger variety of images for ‘visual browsing’ for certain search terms
  • ‘About This Result’ enhancements
  • Shopping using Google Lens
  • In-person shopping at home
  • ‘Things To Know’ - refine and broaden searches

MUM Explained 

Google first showed us a preview of MUM back in June where the AI was used to find variations of covid vaccine names in 50 different languages in just a few seconds. This doesn’t really help paint a picture for how MUM will be applied to everyday search though.

At Google’s Search On, they presented an example where we get to see how Google Lens and MUM can be combined to search for specific queries and return more detailed results.

Examples of this are taking a picture of a bicycle part with Google Lens and combining that search with the text “how to fix”. Google’s MUM is then able to identify what is in the image (bike chain/gears) along with the text query “how to fix”, and combine the two with the understanding and context of what the user's intent is. The search results return a video snippet of how to fix a bike derailleur along with written guides on fixing bikes, view the example below.

MUM Explained 

The second demo is a more e-commerce based example where users might find an image or pattern that they want to buy but aren’t sure which retailers would sell it, or what to search to find said item. This new search format allows for more dynamic queries which users would struggle to find without the addition of combined image and text search. 

The way Google showcased this is with a user searching for an image of a floral shirt but adding text to refine the result to include socks with this pattern. See the example below.

MUM Explained 

I can see this feature being very handy when trying to find results for something you can’t put into words, it would be great from a home DIY point of view or if you are someone who saves a lot of screenshots of clothes or other items and wants to investigate them later using a combination of Google Lens and MUM.

This combo has the potential to return results we have not seen before by merging two different query modals and could create opportunities in search.

Expanded SERP Features Explained

It’s safe to say that MUM stole the spotlight at the Search On event but there are still some notable changes happening to Google’s search results which could impact users differently to the current norm.

Larger Number Of Images For Visual Browsing

This change seems centred around creating more diverse results based on certain keywords and search terms. The example given was a user searching for “pouring paint ideas” and showing results which pull content from the webpage into the search results allowing the user to ‘see’ more of the content before handing over that juicy click attribution.

Larger Number Of Images For Visual Browsing

Google said this interface makes “it easier to visually browse to find what you are looking for.” and I can see where they are coming from. My concern with this kind of result mark-up is that it adds more clutter to the already cluttered search results and pushes more organic results off page one, making the fight for the top result even more competitive. The feature also takes your ever-lucrative content and images off your website and displays this in the SERPs, meaning the user might get the information without clicking to your site adding to the no-click culture.

“About This Result” changes

The addition of ‘about this result’ was injected into search in February and updated in July to give users a snapshot of information on the result as a way to boost trust factors and let the user get an idea of the site without committing to a click. 

The changes they are making include what the website says about themselves which in most cases will be scraped from the ‘about us’ or static information pages held on the site. Google will now also show related results like reviews and what competitors are saying about the same topic. View an example of this below.

“About This Result” changes

Google confirmed that this information is automatically populated and cannot be controlled, meaning there is no markup or section on search console to amend any mistakes or misinformation easily. We have been told that the information is gathered through a set of undisclosed algorithms and that Google will keep an eye on any misuse to stop any foul play. 

In Person Shopping At Home

The idea behind this feature is to allow users to search for shopping items but refine the search in Google allowing the user to create custom filters like in stock, localisation and category type. 

This update to the search results will be launched first across selected markets including U.S. (English), UK, Australia, Austria, Brazil, Canada, Denmark, France, Germany, Japan, Netherlands, New Zealand, Norway, Sweden and Switzerland on September 29th.

In Person Shopping At Home

While I can see some benefits for the user in this functionality, the first thought that springs to my mind is that this adds to the zero click culture Google is creating by adding these result features. Currently the estimated percentage for no-click searches is around 66% which has been a direct cause of rich results and SERP features like local 3-pack, Google Shopping, People Also Ask, PLA’s, Google My Business. 

While Google may be trying to create a better experience for it’s search users, it is inadvertently taking away users from visiting a site and then browsing for desired product/information. While sizable brands might not feel any impact from this, the mid and lower tear business could find themselves being pushed out of search.

‘Things To Know’ - refine and broaden searches

It was also announced that some new versions of SERP features will be added and improved by pairing them with MUM (Multitask Unified Model). 

The example Google gave was a featured snippet panel called ‘Things To Know’ which is very similar to the current ‘People Also Ask’. The feature populates based on the users search query and provides what Google thinks are things you should know relating to the topic which might help you get the information quicker. 

The example Google gave was a user searching for ‘acrylic painting’ and as they scroll down to the ‘Things To Know’ panel it provides sub categories and other popular related searches to try to answer the search query, view this below.

‘Things To Know’ - refine and broaden searches

Google will also be introducing ‘Refine and Broaden searches’ by enabling a feature that takes the user’s query and applies it’s AI to suggest results that might be what you are looking for. 

If we base it on the above example of ‘acrylic painting’, the ‘refinement’ shows more granular topics like ‘acrylic painting techniques’ and the ‘broaden’ feature shows related terms which are less specific but based on what you have searched for, like ‘styles of painting’.

A feature like this already exists at the bottom of Google’s search results page called ‘Related Searches’ where it generates other search queries based on what you have typed. The difference here is that the ‘Refine and Broaden’ feature is injected higher in the search results page, as shown by the example above.

While I have no doubt that myself and many others will end up using this feature from time to time when it pops up based on what I have searched, it doesn’t strike me as a feature that will refine my results for the better, I get more of a sense this will send users down the rabbit hole detracting them from their original query. This is also another rich result which will be prioritised and added above the fold which will push organic results down and add more features to the search results.

What will be the outcome of these changes?

When MUM was first announced and presented to us as a new AI that would be 1,000x more powerful than BERT it was hard to conceptualise, but the recent examples and detail of how this would work in conjunction with search begins to paint a better picture.

The way MUM can be paired with features like Google Lens to not only search images but add queries to generate unique results is very impressive and a feature that is likely to bring about positive change where it is applicable. There are limitless possibilities that can be created from this combination and it may create an opportunity to serve more creative results.

However, some of the new features that expand on existing search results like related topics in videos, things to know, refine this search and broaden this search, look to keep the user on Google’s platform while scraping data, images and snippets from website content and displaying it themselves. This reduces the likelihood that users will click through to your site as the information is displayed in the search results. Tracking users and how they interact with a website is important so that trends and metrics can be analysed to find opportunity and drive performance. 

If the user doesn’t click through to your website because Google has created a monopoly of search result features to display information you have crafted or just selects certain products to display rather than ranges of products which have been merchandised on your website, you will end up missing out on building a good idea of your audience along with the potential loss of a sale. 

It’s clear that Google is making these decisions with the user at the centre of it’s mind but I would love to see some control put back into organic results so it doesn’t have to fight all these markup new features.

Beu Smith

SEO Lead specialising in techical SEO across eCommerce, B2B and B2C. Passionate about delivering organic growth.