Google has published a new documentation on the functionality of research results on AI that explains what it is and how SEO and publishers should react. AI glimps are a new border for the research marketing community, which makes it essential to understand the documentation before forging an opinion on what to do next.
Which triggers the previews of the AI
AI Presentation is a new Google search results feature that offers natural language responses in the form of answers and links to web pages so that users explore the answers more. AI’s overviews show that user’s intention is to quickly understand information, especially when information needs are linked to a task.
“AI glimps appear in Google search results when our systems determine … when you want to quickly understand information from a range of sources, including web information and web information and web information and web information Google knowledge graphic. “”
In another part of the documentation, he meets the trigger for the information -based information needs:
“… and use the information they find to advance their tasks.” “”
What types of sites in AI preview are linked?
An important fact to consider is that simply because IA preview is triggered by the need for a user to quickly understand that something does not mean that requests alone with a need for information trigger the new research feature. Google’s documentation clearly indicates that the types of websites will benefit from the links on AI glimps include “creators” (which involves video creators), electronic commerce stores and other companies. This means that much more than information websites that will benefit from IA previews.
The new documentation lists the types of sites that can receive a link from the IA preview::
“This allows people to dig more deeply and discover a diverse range of content from publishers, creators, retailers, businesses and more, and use the information they find to advance their tasks.”
Where the previews of the sources of information
AI glimps display web and knowledge graphic information. Large languages must currently be completely recycled from the ground when adding large quantities of new data. This means that the websites chosen to be displayed in the functionality of the overviews are selected from the standard Google search index which in turn means that Google can use the generation (RAG) (RAG) of recovery.
RAG is a system that is between a large language model and a database of information external to the LLM. This external database can be specific knowledge like the entire content of HR policies of an organization to a research index. This is an additional source of information that can be used to reveal the information provided by an LLM or to show where to learn more about the question that is answered.
The section cited at the start of the article notes that AI’s overviews cite sources on the web and the graph of knowledge:
“AI glimps appear in Google search results when our systems determine … When you want to quickly understand information from a range of sources, including information on the web and Google knowledge graphic . “
What the automatic inclusion means for SEO
The inclusion in AI previews is automatic and there is nothing specific to the IA previews Seos need to do. Google’s documentation indicates that following their directives for the classification in regular research is all you have to do for the ranking in the IA previews. Google’s “systems” determine which sites are chosen to appear for surfaced subjects in AI previews.
All instructions seem to confirm that new overviews of data index data index functionalities. It is possible that Google Filters the search index specially of IA previews, but deactivated, I cannot think of any reason that Google would do.
All instructions that indicate automatic inclusions indicate the likely possibility that Google uses the regular search index:
“No action is necessary for publishers to benefit from IA’s overviews.”
“AI glimps display links to resources that support information in the snapshot and explore the subject more.”
“… a diverse range of content from publishers, creators, retailers, businesses and more …”
“To classify AI’s overviews, publishers must only follow the Google Search research guide.
“Google systems automatically determine what links appear. There is nothing special to do for creators to be considered as following our regular advice to appear in research, such as the Google Search essentials cover. “”
Think in terms of subjects
Obviously, keywords and synonyms in requests and documents play a role. But in my opinion, they play and an oversized role in SEO. There are many ways that a search engine can annotate a document in order to match a web page to a subject, such as Googler Martin Splitt qualified Annotation of the centerpiece. An annotation of the centerpiece is used by Google to label a web page with what this web page is.
Semantic annotation
This type of annotation links the content of the web page to concepts which in turn give a structure to an unstructured document. Each web page is unstructured data so that search engines must understand this. The semantic annotation is a way of doing so.
Google corresponds to the web pages to concepts since at least 2015. Google web page About their cloud products, explains how they have integrated neuronal correspondence into their search engine in order to annotate the content of the web page with their inherent subjects.
This is what Google says about how it corresponds to the web pages to concepts:
“Google Search began to incorporate semantic research in 2015, with the introduction of remarkable IA research innovations as the classification system in deep learning Rankbrain. This innovation was quickly followed by a neuronal correspondence to improve the precision of the recovery of documents in research. Neuronal correspondence allows a recovery engine to learn the relationships between the intentions of a request and very relevant documents, allowing the search to recognize the context of a request instead of simple similarity research.
Neuronal pairing helps us to understand the blurred representations of concepts in requests and pages and associate them with each other. He examines an entire request or page rather than simple keywords, developing a better understanding of the underlying concepts represented in them. »»
Google has been doing this, corresponding to the web pages to concepts, for almost ten years. Google’s documentation on AI’s overviews also mentions that the display of links to web pages based on subjects is part of the determination of the sites classified in the IA previews.
IA glimps show links to subjects
Here’s how Google explains:
“AI glimps display links to resources that support information in the snapshot and explore the subject more.
… IA’s overviews offer an overview of a subject or a request based on a variety of sources, including web sources. »»
Google’s orientation on subjects has been one thing for a long time and it has gone well that SEOs have attenuated their grip on targeting keywords and also begin to give the subject targeting a chance to enrich their ability to surface Content in Google research, including in AI previews.
Google says that the same optimizations described in their essential research documentation for the ranking in Google Search are the same optimizations to apply to the rank of Google.
This is exactly what the new documentation says:
“There is nothing special to do for creators to be considered as following our regular advice to appear in research, as covered in Google Search Essentials.”
Read the new documentation linked to Google on AI previews
Presentation of AI and your website
Star image by Shutterstock / Piotr Swat