Google Search On: Mum is coming faster than expected

In the “Search On” live stream, Google has announced new features that are to be rolled out in Q4 2021 or Q1 2022. Surprising: These are already the first implementations by Mum.

Google had only announced Mum in May. When the update would take place, there was no more specific time than “in the coming months and years”. At Google’s “Search On” event, the first features have now been presented, in which Mum will at least partially be used in the foreseeable future. There are particularly interesting innovations in the areas of shopping, search and Google Lens.

Google declares that it wants to make the search experience multimedia. This is where Mum, the Multitask Unified Model, an AI that can understand and generate language, helps – better than Bert. Since Mum is multimodal, she should also understand images and, in the future, video and audio files. The first updates should be available in the “coming months”. Then users should be able to combine images and text within a search query: For example, if a picture of a patterned shirt is open, the Google Lens can be opened. With the selected image section and a typed-in search query such as “socks in this pattern”, the image, namely the pattern, and the search query are combined in the form of text. Shopping results are then spat out in the example.

This should be particularly helpful because many people do not always have the right name for everything ready – for example what the special flower pattern is called or what all the individual parts of the bike are called. Google cites this as a second example: Users can take a photo of a broken part on the bike with the Google Lens and type in a query such as “how do I fix it”. The AI ​​should recognize the damaged part, understand the intention based on the search query and display advice or even videos for the repair as results. The Google experts expect the integration of the Google Lens into the Google app in the time window from Q4 to Q1.

Mum also changes the SERPs fundamentally. The first new features were announced in the live stream – however, there is still no time for the roll-out. The first change is basically an expanded and improved version of “Similar Questions” – Mum should be able to identify how users develop topics. So Mum would evaluate which topics are most interesting to seekers first. If someone is looking for “acrylic painting”, for example tutorials and instructions, different styles or how splashes of paint can be brushed away.

There should be the possibility of this to broaden or specify the search. Refining the search query is already known from the related search queries at the end of the SERPs. So far, the principle has been very simple; longtail keywords are more or less simply output here. This box would then not appear at the bottom, but in the middle of the page and probably play out finer search queries. The other possibility of generalizing the search query would then display main topics such as “painting” or “famous painters”, as can be seen in the example:

This is how the boxes for broader or more precise search terms should look. (Image: Google)

Finally, the SERPs snippets will provide additional information about the publisher. For this purpose, on the one hand, own information is used, such as the “About Us” pages or the imprint, but also reviews from other organizations or pages. With this, Google wants to make it easier for users to check sources without great effort. Similar search results should also be displayed there.

The updates in the shopping area will initially only be rolled out in the USA n. Overall, Google wants to create a very visual shopping search. In the future, Google Lens should be able to recognize and display products shown in photos. The Google Lens is activated on a website with photos, on which small dots then appear. These mark where a product was recognized. With a tap on it, shopping results should then be displayed – it looks almost like on Instagram. Products with the Google Lens should also be able to be searched for on the desktop using an image search. The products are displayed on an overlay at the edge.

There are also changes in the SERPs – if products are searched for (such as “Crop Top”), products and stores that sell them are also primarily displayed. Search results that are not products or shops, for example guides, are displayed significantly less cannot be priced in stock. There is no further information here as to how much of this has to be done manually, or to what extent this is or can be automated – i.e. whether this function can be realistically implemented at all. According to Google, this function will be rolled out in the USA and 14 other countries with immediate effect.

Videos that are in the SERPs not directly to a website far but are played in the SERPs, buttons with further topics and search queries will be given in the future. Mum should be able to identify topics in the videos, even if they are not explicitly mentioned. In Google’s example, buttons for further information about penguins, seals and an island are displayed for the video about penguins, as well as various search queries such as “Macaroni penguins resume”. This is not mentioned in the video, but the topic is deduced from the information shown, such as the search for family members and dealing with predators. This update should start in the coming weeks, there should soon be an opt-in option.

In all Google examples, multimedia search results from various Google areas are displayed everywhere: Shopping results in the Google Lens and more pictures, videos and also shopping results in the regular search up to overlays with products or further information. Google seems to be developing a large ecosystem, as separate areas for an image search or a shopping search are probably out of date. This should be more user-friendly: Nobody has to know exactly what the names of the things you are looking for are called – you can search with images and photos and view the results in text, video or image form. The multimedia, made possible by Mum, should also reduce or bypass language barriers. However, it also appears that information is separated from its respective places of publication. Information should be available to users anywhere at any time, which, conversely, also means that they may jump more frequently from site to site. So while it initially looks like e-commerce is winning, it is also much easier for people to compare prices and go to the cheapest provider. For search engine optimization, this will probably also mean significant changes – for example in terms of the relevance of keywords and the required quality of content.