Fewer clicks, more conversations: Google’s AI Mode is here. What now?
From October 8, you can talk to Google in Italy too. Type in a query, launch the search, and you will no longer see the SERP and the list of results: there is a dialog box where you will find the answer, with the sources cited alongside it.
This is AI Mode, the new evolution of the search engine which, like AI Overview, is based on the Gemini model but goes even further, literally covering the entire screen of your device, leaving no room for anything else, at least for now. In the United States, users have been living with it for months, with a mixture of enthusiasm and mistrust. And if you work on the web, the point is obviously different: how much traffic will remain on websites if Google answers for you? Will publishers see further declines? And what will change for e-commerce?
This is not the time to panic, but to understand how to remain readable for a search engine that no longer just indexes, but interacts with the information it finds. And above all, to find a way to achieve visibility, which also involves using the tools that SEOZoom already makes available to you.
What is Google’s AI Mode
AI Mode is Google’s new conversational search mode, powered by the Gemini language model and designed to transform the search experience from a list of results to an interactive dialogue.
We can consider it Mountain View’s strategic and powerful move to counter the rise of chatbots and conquer the new market of complex problem-solving, a direct evolution of the Search Generative Experience (SGE) that users in the United States have been testing since May 2023.
This mode is not enabled by default, but you can find it by selecting the AI Mode button that appears next to the search bar on desktop and mobile devices: Google replaces the traditional display with a full-screen interface, with a dialog box that opens to the conversation flow, which you can continue in the way you find most useful.
Not all queries, however, activate AI Mode. Searches with commercial or transactional intent remain anchored to the traditional SERP for now, where Google can show products, images, and price tabs. AI Mode comes into play on exploratory questions: comparisons, opinions, decisions to be made, emerging topics. That’s where the engine tries to “reason” along with the user, turning the search into a journey, not a click.
From a technical standpoint, AI Mode is based on the same infrastructure that powers AI Overview, but with one key difference: while AI Overview is a static information block within the classic SERP, AI Mode is the search itself, an interactive environment in which the engine conducts a continuous search session, fueled by your requests. It replaces the entire results page, maintaining the ability to click on sources but drastically reducing the immediate visibility of sites. And it goes even further: in the United States, where AI Mode has been active for several months, Google has also begun experimenting with “agent” functions, i.e., the ability to perform actions directly from the interface, such as booking a restaurant or creating a personalized itinerary.
How AI Mode search works
When you try the feature, you immediately realize that you are no longer “searching,” you are talking to Google. The engine returns a complete, pre-processed text, and next to it are the sources that inspired it. There are no blue links in a row, no endless scrolling through SERPs: there is a single screen that summarizes, cites, and updates as you add questions.
The system is designed to analyze and reformulate results using a process called query fan-out: the user’s question is broken down into sub-queries and sent to the engine to search for answers on multiple fronts simultaneously. The information collected is then summarized by Gemini, which produces a coherent version and, where possible, supports it with direct quotes or reference links.
See what happens when you activate it for a complex query such as “Organize a weekend in Rome on a budget of $300, including archaeological sites and typical restaurants.” There is no traditional SERP, but the entire screen is filled with a response generated in real time: a structured itinerary, with suggestions for transportation, museum opening hours, and an estimate of costs. The AI responds in natural language, synthesizing information from multiple web pages and dynamically citing some of the references from which it drew the information.
On the right, you will find a carousel of sources—in this case, links to booking sites, travel blogs, and guides that the AI used to build its response. It is the new SERP, less impactful, perhaps even less visible and less “engaging.”
And below that is a dialog box that says “Ask anything,” allowing you to continue your search on the topic or change the subject completely.
What changes for users and content creators
The direct implications of this system are immediately clear.
You, the user, stop typing a keyword and describe a need – gone are the days when you had to hope to get the exact match right because the algorithms were “robotic”! Today, you dialogue with the engine and it gives you a complete action plan: an itinerary divided by days, with estimated costs for accommodation, food, and transportation, suggestions on what to see, and even tips on how to save money.
And this is the point that revolutionizes the work of content production: Google responds for you. It takes your content, summarizes it, uses it to satisfy the user and, in the best case scenario, gives you a mention in what effectively becomes a vertical “new SERP,” where you appear alongside your competitors. The technology that makes this possible, described by Google engineers as capable of “multistep reasoning” through query fan-out, effectively strips your content of its primary value, which is to attract a direct click.
This is why, especially for publishers and news sites, the arrival of AI Mode raises real concerns about traffic sustainability.
Source selection logic and ranking mechanism
But who appears in AI Mode? Which sites gain visibility in this new mode? According to official statements, the sources displayed come from the public web, selected based on their relevance and authority as perceived by the model. Google claims that it does not draw on paid or restricted content, and that it favors reliable and up-to-date sites, with clear identification of the author and recognizable signs of quality.
From what we know, there are no external databases or confidential sources—for now, Google confirms that AI Mode shares the same infrastructure and draws on the same content archive as the old search engine, but filters it according to different parameters and applies autonomous logic in the final evaluation.
The sources that end up in the AI response must already be indexed, accessible, and deemed reliable by ranking systems. Google refers to “high-quality web content,” i.e., pages that comply with Search guidelines, with clear signs of experience, expertise, and authority.
The new part comes after selection: unlike classic ranking, AI Mode does not sort results, it combines them.
Gemini reads the texts, isolates the relevant passages, and constructs a generative response, where each sentence can come from different sources. It is a double selection that takes place on the same, single Google index.
The first selection is in the normal SERP. When AI Mode breaks down a query with its query fan-out, it launches searches on the traditional index. Imagine again your complex query “Organize a weekend in Rome on a budget of $300, including archaeological sites and typical restaurants”: the system does not search for that exact phrase in its index, but breaks it down into a tree of logical sub-tasks, which are then translated into a series of semantic queries that the machine runs in parallel against the traditional web index.
It is a simplification to think that it searches for the dry keyword “cheap flights to Rome”; its internal queries are more like “average cost of hotels in Trastevere in October” or “Rome metro timetables at the weekend.” AI acts like an ultra-fast human researcher: first it understands your need, then it writes a list of questions to ask Google Search. This has been confirmed by some official company sources: the Gemini model uses the “old” index in a new way to access fresh and authoritative sources.
The AI analyzes the results returned by the classic algorithm, using the usual ranking signals (relevance, PageRank, E-E-A-T, etc.) to decide which documents are the best candidates to “read.” Finally, it writes its summary based on the sources it considers most reliable. This is why traditional SEO is not dying and is, in fact, the essential basis, the fundamental prerequisite for being chosen as a source. Your strategic opportunity is to anticipate this “fan-out” by creating a single, authoritative asset that contains the answers to as many of these micro-queries as possible, to become the backbone of the entire AI-generated response.
The second selection is new. Once it has obtained the list of candidates, the AI stops classifying the URLs and starts “casting” the information within those URLs. It does not look for the best page, but the most useful piece of text, the clearest table, the most reliable data to extract and include in its summary. At this stage, the criteria are structure, clarity, and information density. Your strategy, therefore, has a dual objective: to use SEO to pass the first selection and become a candidate; then, to structure the content to win the final “casting” and be the chosen source for the answer. It is SEO for AI, or GEO if you prefer, that we have been talking about for months.
Instead of a visible link ranking, there is a process of semantic weighting, in which the model gives more weight to content that is clear, consistent, and written in a way that is readable even out of context. This is why many technically well-optimized but dense or redundant pages end up not being cited. The AI Mode ranking is a synthesis, and the more interpretable your content is, the more chance it has of becoming part of the answer. And when the system doesn’t “trust” the result it generates, it reverts to the traditional SERP — a way to avoid inaccurate or unverified answers.
The impact on business: traffic, ADS, and queries
For over twenty years, Google Search has been the “gateway” to the web, a comforting and almost surprising presence in its “banality”: a blank page with a search bar to fill in. AI Mode also starts from here, even if it then revolutionizes (this time the term is not an exaggeration) everything else.
For two decades, the agreement between Google and content producers has been clear, albeit tense: you created quality content, Google indexed it, ranked it, and sent you traffic. Your business model was based on monetizing that traffic through advertising (CPM/CPC) or leads. The entire SEO industry was based on this equation: a good ranking brought visits, and visits generated economic value. This balance was already precarious. Battles over featured snippets, the rise of “zero-click” searches, and disputes with publishers over the appropriation of content in Google News had already shown Google’s tendency to keep users on the SERP.
Now that balance is broken. AI Mode brings that trend to its logical conclusion. It is no longer a matter of “stealing” a fragment for a snippet, but of using the entire corpus of your content as raw material to generate a proprietary response. The engine ceases to be a ‘vehicle’ that brings traffic to your door and becomes the “destination” that responds in your place.
And the numbers coming out of the United States, where these features have been active for months, confirm this expropriation of value. Industry analysis, such as that from Similarweb, shows an average 30% drop in clicks on organic results when an AI block appears. BrightEdge has recorded peaks of 40% in purely informational sectors. This happens because the user finds the answer and no longer has a reason to click.
All is not lost for content producers
However, be careful about declaring “the end of the web” or other such dire messages.
First of all, as mentioned, not all queries trigger this scenario, and in particular, searches with obvious transactional or commercial intent remain “sheltered” from the feature (but the SERP is still cluttered with all the other Google features that hide organic results!).
And then, as we saw with the initial data on AI Overview: it is true that there is a drop in traffic—which has a direct impact on a business model based on visit volume—but the pages mentioned show an increase in “brand recall” and trust. The user does not immediately open the site, but recognizes it as a source. It is a change in perception that matters, especially in a context where digital reputation is as important as traffic. It is a requalification of the click: you may receive fewer visits, but from more informed users with a stronger intent.
Finally, there is the human factor: trust. The adoption of AI Mode is not a given. Several studies and surveys conducted in the United States show that a significant portion of users find AI responses unreliable, too wordy, or simply wrong. Many say they would prefer an option to disable the feature and return to the classic SERP. This skepticism means that users will still seek direct validation from original sources.
In either case, what matters is that you need to be found.
The future of ADS: not the end, but a change
At present, however, there is still a sore point: Google Ads, a $180 billion a year business that cannot disappear—and which Google cannot and will not give up, or it would be corporate suicide.
Google’s challenge is to shift the point of monetization from clicking on the link to interacting within the response; they are already testing and implementing various solutions, and there could be three possible forms, which make the ad a native and almost indispensable component of the solution provided by AI:
- Sponsored links in the response. The most obvious route. In the AI-generated response, some of the source links cited or products suggested will be marked as “sponsored.” Imagine a travel itinerary where the recommended hotel is a paid placement—the ad ceases to be an external banner and becomes an integral part of the response.
- Integration with Shopping Graph. For all queries with commercial intent, AI responses will be populated directly from Google Merchant Center data. You will see carousels of sponsored products, price comparisons, and “buy now” buttons that lead to paid product listings. AI becomes a personal shopper, and its recommendations are monetized.
- Paid actions and lead generation. For service queries (“find a plumber near me”), the AI response will not be text, but an action: a sponsored “Request a Quote” or “Book an Appointment” button, which generates a paid lead for the advertiser.
The differences between AI Mode and AI Overview
Now you may be wondering: wasn’t AI already in Google? Why another feature? This question had already arisen in the analysis of Google I/O 2025, when the company’s CEO Sundar Pichai presented the sensational feature. And perhaps today we have the answers.
First of all, AI Overview and AI Mode are two different features that come from the same engine and are, in fact, interconnected in a two-stage process.
The AI Overview is a box within the SERP: a summary that appears above the results. It is a direct evolution of the featured snippet, a steroid injection for the traditional SERP. It appears automatically for queries that require a summary of multiple sources, such as procedural ones (how to remove permanent marker from wood) or comparative ones (what is the difference between a cappuccino and a caffè latte). Its purpose is efficiency: to give you a quick and structured answer. But it’s not just a point of arrival: at the end of the box, you’ll always find the old SERP, but also a call to action that invites you to “Learn more in AI Mode,” transforming the Overview into a gateway to the next experience.
By clicking there, or activating it directly, you enter AI Mode and the interface is transformed. The function takes over the entire page, the SERP disappears, and only the dialogue box remains, with the AI’s answers and a list of sources that updates as you continue the conversation. It is a separate environment that the user chooses for a different intention: they are no longer just looking for efficiency, but for in-depth information. They enter here to solve complex problems, to plan, to dialogue with information.
Comparing the answers: two queries, two castings
A comparative analysis of the results of AI Mode and AI Overview in relation to the same query gives you definitive proof that you are not looking at the same SERP.
Let’s start again with “how to remove permanent marker from wood.”
If you search using traditional search, Google activates an AI Overview; the output is an efficient box, inserted above the blue links, which gives you a list of quick methods (“Denatured alcohol,” “Baking soda paste”) and “Important tips”—it’s a quick answer. And the sources? The carousel at the top shows sites such as “idealista,” “homify,” and “immobiliare.it,” in a selection of mainly Italian candidates—and, it is worth noting, not always from the top three in traditional rankings.
If you activate AI Mode for the same query, the interface changes completely. The output takes up the entire screen and is no longer a list, but a complex procedural guide, which divides the methods by context (“Methods for finished wood” vs. “Methods for untreated wood”) and adds a six-step “General Procedure.” And the sources are global “how-to” giants, starting with UGC platforms such as Reddit, Quora, and Wikihow, with links to the English versions.
This is no coincidence, because the pattern repeats with other queries.
For example, in “What is the difference between cappuccino and a latte?” the AI Overview shows a summary table and cites authoritative websites, mainly Italian (kitchenaid.it, cookist.it), as sources.
AI Mode also expands the response here, with a visibly richer and more detailed table, with more rows and more context, and above all, it draws on an international pool (thecoffeebean.com, nescafe.com).
The implications for your strategy
These simple examples alone give you at least three operational guidelines for your content strategy.
First, the functions operate with two separate “castings”—it is not enough to optimize only for the Overview and hope to “win” in AI Mode. Second, competition in AI Mode is global: you are no longer competing only with sites in your market, but with the best information resources on that topic worldwide, even in other languages. Third, depth pays off.
To intercept the AI Overview, your tactics must be surgical: you need “atomic” content, i.e., single pills of knowledge, precise data, and blunt answers. To become a source for AI Mode, your strategy must be holistic: instead of “atoms,” you need to build complex “molecules,” such as complete procedural guides and comparative analyses that map and solve an entire user need.
Why two functions? Google’s strategy
Returning to the question of the “logic” behind Mountain View’s moves, the answer is not technical but strategic, and concerns the very survival of Google’s business model in a world invaded by conversational artificial intelligence. AI Overview and AI Mode are not two versions of the same thing, but different weapons to fight on two fronts.
AI Overview is a defensive move. Its task is to protect the existing classic search business from the threat of competitors such as Perplexity. By integrating a summary directly into the SERP, Google makes traditional search more efficient, reducing the incentive for users to go elsewhere for simple questions. The goal is to fortify the castle, keeping users within the ecosystem for the searches they are already accustomed to.
AI Mode is an offensive move. Its task is to conquer the new market of complex problem-solving, responding to the existential threat posed by ChatGPT. Google has realized that when it comes to planning a trip or comparing complex strategies, the “list of 10 links” is becoming obsolete. Users want the solution, not the ingredients to build it themselves. AI Mode is Google’s attempt to become the go-to platform for this new behavior before others do.
The SEOZoom operational framework: 3 guidelines for action
What we know about AI Mode and what we have analyzed from the US leads to a single operational conclusion: your SEO strategy must evolve to win the “double casting,” reinforcing the work you had already started to gain visibility in AI Overview.
Your first goal is to pass the classic algorithmic selection to enter the shortlist, and the second is to be chosen by the AI as the source for the final summary. You need to work on optimization for extraction by applying three specific guidelines, which are already at the heart of SEOZoom’s methodology and tools: govern the entire topic, structure the content for the machine, and transform your brand into an unassailable source.
- Govern the topic, not the keyword
Optimization for individual keywords is an approach that has been superseded by the query fan-out mechanism. If artificial intelligence breaks down a complex need into dozens of micro-queries to find the best answer, your content must anticipate this process. You need to stop focusing on individual keywords and start governing the entire topic. Use Keyword Infinity and Question Explorer to map the semantic universe surrounding a need, identifying all related concepts and questions. Your goal is to build pillar content, complex “molecules” that cover the topic so comprehensively that they become the most efficient source for AI during its assembly phase.
- Structure for extraction, not just for reading
Now you have another reader, after the human one and the classic Google search algorithm: the machine, whose purpose is not to read, but to extract. To pass the “second casting” and be chosen as a source, your content must be formatted for algorithmic “parsing.” You need to go beyond human readability and structure the information to make it “plunderable.” Use HTML tables for comparative data, provide clear and concise definitions for key entities, and organize procedural guides with numbered lists. Build FAQ sections based on real user questions found in Question Explorer. To validate this process, use SEOZoom’s AI Engine to analyze the relevance and structure of your text by simulating the perspective of an algorithm, with the necessary additions to make it a flawless data source.
- Turn your brand into a source (E-E-A-T)
Being cited by AI is an act of trust. When multiple sources provide similar factual information, the principles of Experience, Expertise, Authoritativeness, and Trustworthiness (E-E-A-T) become the decisive selection criteria. This is your main strategic asset. Work scientifically to increase your domain’s Zoom Authority, which measures your thematic authority. Build unassailable author profiles that demonstrate real, verifiable expertise. Get mentions and backlinks from relevant sources that act as a vote of confidence. Encourage and value reviews and discussions from your community to build the “social proof” that AI actively seeks. The goal is to transform your brand from a commercial entity into a trusted knowledge infrastructure.
