SGE: Google's answer to Bing's ChatGPT integration in search

SGE is Google's answer to the growing importance of AI in search. The concept was presented at the Google I/O 2023 conference and has since been known under the project name “Magi”. In contrast to Bard, a chatbot, SGE is an AI-based search that aims to transform Google from a pure search engine into an answer engine. As SGE is still in the test phase, it is currently only being made available via the Google Search Lab. So far, SGE is available in the USA, India and Japan.

The idea behind it is that users receive AI-generated answers to their search queries directly, without having to click through the multitude of pages provided by Google. Google is trying to make searches more visual, personal and interactive. Snapshots give users a quick overview of the topic - including sources and further links. SGE uses language models such as MUM and PaLM2 and is particularly sensitive to YMYL (Your Money Your Life) topics. Ads continue to be labeled in a differentiated manner.

AI-generated snapshots: The snapshot provides an overview of various relevant information on the topic - including their sources. The links are displayed as a kind of carousel and can also be accessed via the arrow symbol. Images and explanatory videos are also possible answers to the Google search query.

Clicking on “Ask a follow up” opens the conversation chat with an input field. Possible follow-up questions to the initial search query are also suggested at the bottom of the snapshot in order to delve deeper into the topic. In this way, the search journey is structured like a conversation.

Vertical content: Vertical search content such as Shopping ads or local search results are displayed directly in the snapshot. The Google Shopping Graph displays AI-based results for the search query. The information comes not only from product detail pages, but also from reviews and other sources. The AI compares different results with each other and tries to prepare and display the best result for the search query.

There is currently no official information on SGE, apart from that provided by Google in a PDF. The SBU results are generated by the advanced versions of MUM and PaLM2 in conjunction with the core ranking system. The quality of the outputs is improved by independent search quality raters and targeted human analysis. Red-teaming is used for counter-testing to identify weaknesses in the system and to continuously improve Google's SGE. SGE snapshots are based on information from the entire web and the Google Knowledge Graph.

If content is blocked for the Googlebot using the snippet controls, the AI cannot use this as a link source for the SBU snapshot and, as a result, the content of the page is not displayed in the snapshot.

In the future, providers and SEO tools could also enable the tracking of SBUs in the Search Console. Currently, no usable data is available in the Search Console or other tracking tools, meaning that targeted measurement of Google SGE is not possible.

SGE snapshots cause the organic results to slide further down and be less visible. This can affect both organic traffic and the (pixel) ranking. Google is currently testing various snapshot sizes that push the organic search results down to different degrees. The light version of the AI-generated answers is shorter and only opens fully after clicking on “Show more”. In addition, a distinction is made between search queries with AI-generated answers and without AI-generated snapshots - not every query appears with an SBU answer.

Google is also testing AI-supported image generation with SGE, which harbors new opportunities and risks. Four different AI-generated images are suggested. It is possible that Google will also integrate this function directly into Google Images in order to provide users with customized inspiration.

It is to be expected that the number of zero clicks will increase if an SBU search result provides the right answer right at the start. Similar effects have already been observed with the introduction of featured snippets.

Simple definitions and FAQ content could generate less traffic, as the need for information is already satisfied by the AI in the snapshot. The AI currently still has problems with complex search queries. One solution for SEOs could therefore be to focus on more complex search queries for which there is no “one” answer.

Traditional metrics could also lose relevance, similar to core web vitals. New ranking factors could also emerge, because according to Google, SBU should help users to find helpful information more quickly and easily and serve as a starting point for further research.

SEOs should focus on topics that have a direct impact on search results and that they can influence directly. This includes focusing on structured data, introducing interactive elements to websites, optimizing snippets, mobile optimization and answering frequently asked questions in natural language. The aim should be to provide high-quality content.

Link carousels in the SBU snapshot could become just as important as long-tail search queries. The focus should be on niche information that cannot be answered with a short AI-generated result in order to continue to generate significant organic traffic from the search engine in the future. Google's E-E-A-T signals could become more important in order to illustrate the traceability of expertise and personal experience.

The introduction of SGE in the EU, especially in Germany, has been delayed due to extensive testing, learning phases and regulatory requirements. As a reminder, the launch of Google BARD in the European market has been delayed by five months due to EU regulations. However, SEOs can keep a close eye on the American search market and prepare accordingly. The future of the Google SBU is written in the stars. Stay tuned!

Image source: Google Blog

• Fig. 1: Snapshot in response to a question asked

• Fig. 2: Video as an answer to the question asked

• Fig. 3: “Ask a follow up” including follow-up questions

• Fig. 4: Google Shopping Graph Fig.

• Fig. 5: AI-generated images in Google SGE

Abonnieren Sie unseren Newsletter, um immer über die aktuellsten Updates und Digital-Trends informiert zu sein.