How does the AI Overviews ranking work?

First, here are a few terms to help you understand how Google generates responses to user queries in the AI Overviews section.

1. Fan-out queries

The AIO mechanism begins by breaking the user’s query into its constituent parts, known as “fan-out queries.” This involves analyzing the prompt entered and breaking it down into related phrases. How does it work in practice? You can check it out with this tool, which will show associated queries after you enter a sample prompt: https://queryfanout.ai/. Let’s say a user asks Google, “What are the best SEO agencies in Poland?” The breakdown of phrases for such a query might look like this:

Source: queryfanout.ai

In a nutshell, Google first tries to match a more complex query to related keywords.

In the next step, for these queries, the algorithm queries the Google index to retrieve data and broader context. From an exhaustive list of available sources, the system selects the most meaningful “chunks,” or fragments of text, that it can directly use to construct a response. The goal is to select sources from this exhaustive list that complement each other rather than duplicate information, which is why unique content is particularly important here.

2. RAG

The second key concept in the context of AI Overviews is RAG. This is a technique widely used in LLM tools that aims to reduce hallucinations. The idea is that the model should not “make up” answers, but should rely on reliable sources and provide answers that are as correct and contextually relevant to the query as possible.

How does it work? The RAG system is a connection between the AI model and a database – in the case of AI Overviews and AI Mode, this “database” is the Google index. In this database, we place documents that we want the model to treat as a source of knowledge for generating answers.

3. Grouding

The third concept that helps to understand how AIO works is grounding. Again, this is a mechanism designed to reduce hallucinations, but it works later. Once the model has generated a response, it reexamines the result and compares it to the content of the sources. On this basis, it verifies that the response it has prepared does not deviate from the source material in terms of content.

Do AI Overviews always appear at the top of search results? The answer is no! The layout of the results page depends on what Google considers most valuable to the user in the context of a given query. Sometimes an ad will be higher up, sometimes a map with Google Business Profiles, and sometimes video results.

The part that will be most important to many companies, i.e., the list of cited sources, will not always be clearly visible.

The most common variant is AIO: cited sources on the right, ads below, and only then the good old organic results.

Another option is video results. This layout is especially common for phrases such as “how to do something,” i.e., those where a tutorial recorded in video form may be most useful to the user. In this layout, AIO results appear only below the video and the People Also Ask sections.

According to a report from Senuto’s AI Overviews analysis, Know Simple queries, i.e., simple informational queries, most often have an AIO result, accounting for 57.82% of the cases analyzed. It is phrases with this intent that most often appear in this section, compared to other types. Other industry sources also confirm this.

This means that AIO will most often apply to typical “knowledge” phrases intended to quickly provide the user with information on a given topic. For companies that previously relied on monetizing organic traffic acquired on such phrases, this means they will have to change their business model.

What does this mean for brands? AI Overviews should not be thought of in the context of traffic acquisition. As mentioned above, organic traffic to websites has been declining since the introduction of this solution, because users get the answer on a plate and no longer need to visit the website. In the context of AIO, presence will be significant mainly in the form of mentions (e.g., brand recommendations, graphics displaying logos, and citations as a reliable source of information), and only secondarily in the form of citations (i.e., links to the website).

• A well-thought-out content strategy – in the context of the aforementioned fan-out queries mechanism, i.e., breaking down the prompt into related keywords, it is important to build thematic clusters rather than individual entries. Industry issues should be covered across all aspects, and it is also worth remembering to use internal linking between thematically related content.
• Information gain – Google likes to choose sources that bring something new to AIO responses. Rewriting the content of competitors’ articles will not help you get quoted. Therefore, enrich your content with data, statistics, case studies, and your own conclusions.
• New content formats – the aforementioned example of tap repair shows that sometimes video results win out over standard text on a website. In addition to a blog, consider a YouTube channel.
• Infographics and photos – graphics often appear at the top of AI Overviews results, and this is your chance to get noticed by users. Create valuable infographics with your logo to build brand recognition.
• Logical content structure – in the articles themselves, try to focus only on the topic at hand, write specifically, and remember to use a structure with headings, short paragraphs, and bulleted lists.

Now that you know how the AIO response mechanism works, what phrases it appears for, and what can help you appear in them, it’s time to verify your actions. How can you monitor visibility in AIO?

At the moment, there is no official solution from Google – the introduction of data differentiation to Google Search Console has been announced, but with no specific dates. It is also worth noting that monitoring visibility in AIO poses many challenges, as results are personalized and responses to the same query may vary by user.

What are the options at the moment? Many popular SEO tools offer visibility monitoring in AIO. These include Ahrefs, Senuto, and Semrush. Another interesting option is ZipTieDev, which allows you to track not only citations, but also brand mentions and sentiment, i.e., whether the brand is mentioned in a negative or positive light.

• • The AIO mechanism first breaks the query into phrases (fan-out queries), then retrieves key fragments from the Google index (RAG), and verifies the answer against sources (grounding).
  • It is most often displayed for simple informational queries.
  • How to increase your chances of appearing in AIO:
    • create unique content and thematic clusters,
    • add your own data, conclusions, statistics (information gain),
    • use videos, infographics, and graphics with your logo,
    • ensure a logical, clear structure.
  • You can monitor your visibility in AIO using tools such as Ahrefs, Senuto, Semrush, and ZipTieDev.

At Delante, we constantly analyze AI Overviews results and help our clients position themselves in AI tools using individual AISO (AI SEO) strategies. If you want your brand to be present in AI tools, contact us!

Sources:

https://insight.senuto.com/aio-report/