How much traffic and opportunities is AI Overview taking away from you?

How much do you really know about AI Overview? If your work revolves around online visibility, you know that today’s search landscape has changed, perhaps forever, and much depends on this feature, which redesigns Google’s SERP with a speed and complexity that require clear and constant analysis.

Perhaps you’ve already noticed a drop in organic traffic, an ever-widening gap between impressions and actual clicks, and you’re wondering how to interpret this data and, above all, how to act. There is a lot of confusion, information is fragmented, and the real impact is often difficult to quantify.

That’s why we’ve decided to take stock of the situation and bring you straight to the heart of the change: official figures, practical analysis, and concrete strategies for tackling the new era of AI Overviews on Google.

AI Overview changes Search: already 2 billion users per month

In the latest conference call with investors, Sundar Pichai left no room for doubt: AI Overview has become the real star of Google’s strategy. Launched globally between March and April 2025, the feature now reaches over 2 billion users every month, spread across more than 200 countries and active in 40 different languages.

Protect your visibility!

Check now if your presence in AI Overviews is truly effective and where you can improve.

Find out more

These figures not only confirm accelerated growth compared to last quarter’s already impressive 1.5 billion users, but also mark a turning point for Google’s role in online search.

The Mountain View giant has accompanied these figures with an upward revision of its investment in AI infrastructure, which will rise to $85 billion by 2025. The stated goal? To support the growing demand for AI-based services and consolidate a platform that today effectively “retains” more and more users within its ecosystem.

What you need to know: the different faces of AI in SERP

But let’s move on to what affects you (and us) more closely, namely the implications for those who publish content online. And the question: is AI Overview really to blame for making websites increasingly useless?

First of all, we need to know our “enemy”: AI Overview is not always the same block, nor does it always appear in the same way or place.

Although our minds immediately jump to the classic colored box that appears at the top of the results page—which is certainly its most common manifestation—the phenomenon is actually much more complex and variable.

This feature changes depending on the maturity and stability of the answers Google can give users. Sometimes the summary is presented in a predetermined form, ready to use and identical for anyone doing the same search: here, Google has identified a solution it considers reliable and displays it without hesitation, rewarding the sources that appear in it with constant visibility. In other cases, however, the answer is updated with each search, reworking the content almost in real time and modifying sources and citations based on signals such as location, history, or new questions asked by users.

This alternation between static and dynamic modes is not random: it reflects Google’s need to balance timeliness and authority, providing concise answers when the query allows it and maintaining a flexible approach where the information is still “live.”

Anatomy of an AI Overview: what it looks like and where it gets its sources

An AI Overview is generated by advanced language models and is visually designed to stand out. At first glance, it looks like a block of concise text that aims to answer the user’s question directly, but it is actually a mosaic of dynamically assembled information, with a two-level source attribution system.

It usually consists of

• Summary text: a paragraph that distills the key information.
• Bulleted or numbered lists: for answers that require a sequence or a list.
• Links to sources: a crucial feature, the links are not presented as in a classic snippet, but are integrated into the text or displayed in “carousels” or boxes. By clicking on these, the user should be able to explore the information in more depth.

You can think of it as a research dossier: there is a summary text (the executive summary) and then there are the sources, presented in two distinct ways and with two different roles.

1. The list of overall sources (the “absolute SERP”)

At the end or side of the AI-generated text, Google almost always presents a carousel or grid of cards. Each of these cards corresponds to a web page that the algorithm has deemed an authoritative and relevant source for building the answer as a whole.

Consider this list as the “absolute SERP” of the AI Overview. It is the set of all sources that Google has valued and drawn from to formulate the summary. Being on this list is the first, fundamental level of visibility: it means that Google has recognized your site as one of the most important voices on the topic.

And speaking of sources, Google draws on extremely varied references, but recent studies indicate a predominance of sites with very high authority or with huge amounts of user-generated content. Wikipedia, YouTube, and Reddit alone account for a significant portion of citations, highlighting the algorithm’s preference for established and conversational sources.

2. Specific attribution links (the “paragraph SERP”)

Here lies the real strategic innovation. If you look closely at the text of the Overview, you will notice that specific phrases, data, or paragraphs are often accompanied by a small chain icon. By clicking on it, the user sees only the specific sources from which that piece of information was taken.

This is the “paragraph SERP.” It is a much narrower and more targeted selection than the overall list. Google, in essence, is not just saying, “I used these 10 sites to write this text,” but rather, “To write this specific sentence, I mainly used these 1 or 2 sites because they are the most relevant to this micro-topic.”

The strategic implication: multiple visibility

This two-tier structure has a powerful implication: a single website can appear multiple times within the same AI Overview, even with different pages.

Let’s take a concrete example. For the query “organize a trip to Japan,” the AI Overview presents:

• A first paragraph on how to define the itinerary and the ideal duration of the trip, indicating that “for a complete immersion, a minimum of 15 days is recommended, while 5-7 days may be enough to explore Tokyo,” citing your in-depth guide on stops and timing as the source.
• A second block dedicated to bookings, explaining strategies for finding cheap flights, booking hotels or ryokans, evaluating the Japan Rail Pass, and booking special experiences. In this case, Google can draw on several pages of your site dedicated to transportation, accommodation, and tickets in Japan.
• A bulleted list of essential documents and preparations (passport, insurance, SIM card, adapter, cash), citing your detailed checklist of what to bring and how to get organized before departure as the source.
• A paragraph on costs and budget management, with estimates of typical expenses for a week and practical tips on how to save money, again attributing the information to the most specific page on your site.
• A series of practical tips, such as useful apps to download, basic phrases in Japanese, and dishes to try, for which Google can directly cite your article on travel apps or local cuisine experiences.

In this scenario, your site could appear several times as a specific source within the text, on different aspects of the same overview—from planning to itinerary, from documents to practical tips—as well as being present in the final “absolute SERP.” This is what happens to Sviaggiando, which you can find in the paragraph on “classic itineraries,” in the one on “accommodation,” and in the section on “preparations.”

This shows that the winning strategy is not just to be generic on the topic of “Japan,” but to build vertical, detailed, and authoritative content on every single sub-topic useful to travelers. Google rewards those who can best answer not only the main question, but all the micro-questions that make it up.

The dual nature: the “static” overview versus the “dynamic” overview

But there is another aspect you need to be aware of: one of the most important strategic distinctions we have observed is between the ‘static’ and “dynamic” nature of responses.

• The dynamic AI Overview is what you see generated almost in real time, especially for complex, new, or current queries. The algorithm is actively searching, comparing, and assembling sources. In this state, answers can vary greatly from one search to another.
• The static AI Overview, on the other hand, occurs when Google, after analyzing a huge amount of data, identifies an answer that it considers stable, reliable, and universally valid for a given query. At this point, the answer is “frozen” and served in a much more consistent manner.

The static version is provided by Google when the engine has reached an “accepted” summary for a given query: the answer appears immediately, the same for everyone, and the links to the sources remain unchanged for an often extended period. At this stage, the presence among the sources guarantees stable and predictable visibility.

On the contrary, dynamic mode occurs when Google is still collecting and interpreting data: the response is composed before your eyes, can change with each search, and the list of sources cited is constantly updated. Here, the result is much more volatile and the ranking can vary several times in the same day, following signals of current events, location, and user behavior.

When and why does Google switch from dynamic to static (and vice versa)? There is no public rule, but if you look at the SERP on a “hot” query, you will often notice responses that are slowly being composed, piece by piece. This is a sign that the Overview is still in the dynamic phase. When the answer appears immediately, perfectly formatted and identical to the one from the day before, it has most likely stabilized. This stability can last for days, weeks… then, if user needs change, the Overview becomes dynamic again and starts all over. And the links to the sources change again too.

Why you should really care about this distinction

Understanding which phase you are in is not a technical curiosity, but a business factor.

Entering a static Overview allows you to consolidate your presence in SERPs and benefit from a more regular traffic flow, but it requires constant attention because the situation can change at any time. If, on the other hand, your site is involved in a dynamic Overview, you need to be prepared for sharp fluctuations in visibility and daily competition to be among the selected sources.

But be careful: the game is not over, because Google can “reactivate” the dynamic phase at any time, and your content can disappear from one day to the next.

Knowing how to identify the phase your query is in helps you manage expectations and plan more effectively for content monitoring and optimization to respond to the new logic of generative search.

How fan-out works (and why it can save your traffic)

This mechanism also depends on the completely different principle by which Google analyzes and ranks content, called fan-out.

When a user performs a search, the engine no longer simply interprets the main query and finds the most relevant answer to the literal question, but expands the search to dozens – sometimes hundreds – of related keywords, probing the entire semantic range related to the user’s intent.

Through AI systems such as Gemini and Retrieval-Augmented Generation techniques, Google collects results from a multitude of sources and aggregates them into a summary that best responds to the query.

This paradigm shift renders the “one keyword, one page” logic obsolete: if you limit yourself to optimizing content for only one central keyword, you risk ending up off the radar. Instead, working on the entire spectrum of related intents, questions, and variations becomes the winning strategy.

SEOZoom’s experience proves it: content that responds comprehensively and deeply to all possible variations of a search is much more likely to be selected and cited in the Overview, ensuring visibility even when competition is fierce.

Where AI Overview really appears: not just position zero

The last important thing you need to know is where the box appears. Don’t be fooled by the classic presence of AI Overview at the top of the page, where the featured snippet used to be, because Google is experimenting with new configurations on an increasingly large scale.

Although “position zero” remains predominant as an immediate response when opening the SERP, you can also find AI Overview lower down, integrated among the organic results, especially for queries that require comparisons, detailed explanations, or insights—this is what happens when you search for “organize a trip to Japan,” as you saw earlier. In that case (and in several others), you’ll find the AI box halfway down the page, after the first organic results, and it serves more as an “insight” than a straight answer. On other occasions, the AI box alternates or appears alongside other features—such as “People Also Ask” panels, video carousels, or thematic widgets—creating a results page that is even more fragmented and unpredictable than in the past.

This variety of placements makes any prediction difficult: your visibility may depend not only on the quality of your content, but also on how Google decides to structure the SERP based on the type of query, device, or user habits. The only constant is the need to monitor all possible presentation modes, adapting optimization to an increasingly dynamic and multifaceted environment.

The real impact on traffic: Google rejoices, but the web slows down

While the figures on the real impact of AI Overview are a source of pride for the company, they are a source of serious concern for those working in the world of content and SEO.

For months, studies and analyses have been chasing each other on the decline in traffic to sites caused by AI Overview, and even the recent report by the Pew Research Center in July 2025 leaves no room for doubt: the presence of an AI box in SERP almost halves the click-through rate on organic links.

To be precise, the average CTR on queries that trigger an AI Overview has fallen from 15% to just 8%. The situation is even more drastic for links within the Overview itself: they are clicked on in only 1% of cases, while 99% of users simply read the summary. So, despite Google’s reassurances and the “hope” that users would use the links within the overview to dig deeper, the study shows that users trust (or are satisfied with) the summary and rarely feel the need to check the source directly.

The result is a further explosion of so-called “zero-click searches”: more and more often, the session ends almost entirely on the search results page, without the user visiting any other external sites. In short, people who view an AI response are much more likely to consider their search complete: the AI Overview satisfies superficial informational intent, reinforcing Google’s evolution as a response engine.

Not all searches are equally exposed to the Overview effect. The most noticeable impact is on informative and defining queries: anything that starts with “what is,” “when,” “who,” “how much,” “why” is easily summarized by Gemini and offered directly in SERP. For these intents, the classic “top spot” in organic search has now lost much of its historical value, because users are satisfied with the AI response and rarely continue browsing. The situation is different, at least for now, for more complex queries: comparisons, complex questions, and transactional searches still show a greater propensity to click, leaving room for maneuver for those who provide in-depth content, guides, and comparisons.

These are broadly the same insights we highlighted in our white paper – which you can download (in italian) here: over 11% of Italian searches today activate an AI Overview, with privileged visibility for authoritative brands and vertical portals (the top 5 most linked domains are Wikipedia, Mypersonaltrainer, Msdmanualsm, Santagostino, and YouTube), with strong EEAT. Furthermore, 40% of the sources cited in the overviews do not match the top 10 organic results, opening up new opportunities and unexpected competitors. The sectors most affected are health, knowledge, and how-to, where AI aggregates answers from multiple sources and redefines the rules of online visibility.

The economic implications: who wins and who loses

These figures are not just numbers, but describe a change in the economic model of the web. For years, publishers, blogs, and companies have based their sustainability on a traffic model: creating quality content to attract visitors and then monetizing this attention through advertising, subscriptions, or product sales.

By interposing itself between the user and the site, AI Overview undermines this model. The big winners appear to be Google itself, which keeps users on its platform, and mega-content sources such as Wikipedia, YouTube, and Reddit. The losers are all other publishers—especially small publishers, vertical blogs, and specialized sites—who see their exposure decline dramatically. In fact, for them, the situation is a lose-lose scenario: not only does incoming traffic drop, often by double digits, and with it the opportunities for monetization through advertising, subscriptions, or direct sales, but at the same time, they see their content “used” to train and feed AI, which in return gives them a fraction of the traffic they used to get.

AI Overview in the antitrust spotlight: publishers’ complaint

A “counter-move” by European publishers was therefore inevitable, as they could not remain immobile and indifferent in the face of global success—and the consequences described above.

In recent weeks, the feature has been at the center of a new antitrust complaint filed with the European Commission by the Independent Publishers Alliance, a group of dozens of independent media outlets based in the UK. Google is accused of distorting competition in the online news market through AI Overview, which, by summarizing and reworking publishers’ content, drastically reduces traffic, readership, and, consequently, advertising revenue for news outlets.

According to the complaint, the feature violates the principles of fairness of the Digital Markets Act: Google “uses and reformulates third-party news content to provide direct answers to users, retaining traffic on its own platform and undermining the business model of publishers themselves,” as stated in the document submitted to Brussels.

The European Commission has confirmed that it has launched an investigation and asked Google for clarification on its source selection mechanisms, the transparency of its overviews, and the possibility for publishers to opt out of the AI box. Meanwhile, pressure is growing to find fairer revenue-sharing models and control tools for those who produce original information.

Google, for its part, rejects the accusations, arguing that the overviews “create new visibility opportunities for authoritative sources” and promises greater collaboration with the industry.

How to (really) measure the impact of AI Overview on your site

Faced with this scenario, the most pressing operational question is “how can I measure the impact of AI Overview on my site?” The bad news is that Google does not currently provide a direct tool to do this. The good news is that, with a methodical approach and the help of external tools, you can get a very accurate estimate.

On the Google front, neither Analytics nor Search Console allow you to directly isolate clicks coming from the box generated by artificial intelligence. All traffic from traditional results and from Overview is aggregated under “organic search,” with no specific distinctions. This means that if you limit yourself to standard reports, you risk not seeing—or underestimating—the concrete impact of the new AI responses on your performance.

The most effective way to obtain useful data today is to adopt a comparative approach. Start by isolating informative keywords (those that most easily trigger AI Overview: “what is,” “how to,” “guide,” “when,” etc.) in your Search Console reports. Compare two distinct periods: one prior to the massive rollout of the feature in Italy (e.g., January-February 2025) and one after (June-July 2025). Analyze the CTR delta, i.e., the difference between click-through rates for the same number of impressions and average position. If you notice a sudden drop, especially on keywords that have remained stable in terms of ranking and volume, it is very likely that the presence of overviews is the cause. A further manual check may be useful: search for the queries that have lost the most and check whether Google consistently displays the AI response at the top of the page. This helps you quantify, with a good degree of certainty, how much traffic you have “lost” due to the new dynamics.

This is where the added value of SEOZoom comes into play, as it is currently the only tool that allows you to:

• Accurately and continuously identify all keywords, pages, and domains mentioned in Google’s AI Overviews.
• Estimate the potential traffic associated with these mentions, going beyond simple “presence” and analyzing the real potential for visibility.
• Analyze the type and frequency of mentions and their role within the overviews.
• Measure the overall relevance of your presence in AI Overviews using the proprietary “AI Rank” metric, which assigns a score to the quality and quantity of mentions obtained, helping you evaluate and monitor the actual competitive value of your site in this new context.

By cross-referencing this information with traffic and ranking data, you get a detailed map of the impact on your site and can react more accurately by updating content, strategies, and priorities.

Strategies to stay visible and not lose ground

Faced with a SERP revolutionized by artificial intelligence, the strategy for continuing to generate traffic can no longer be the same as it once was. Today, you need to focus on fan-out: don’t limit yourself to the main keyword, but develop content that captures all the questions, variations, and sub-intentions related to your topic. AI Overview works by expanding the search across a range of related topics: those who manage to cover this semantic spectrum in depth will be much more likely to be selected as a source.

At the same time, quality makes all the difference. Google favors complete, authoritative, and well-referenced answers, penalizing generic or repetitive content. It is essential to update often, monitor your positions, and intervene quickly if you lose ground: the volatility of overviews requires constant attention. If you notice that a page is being overtaken by a competitor, enrich the text, add data, or respond better to emerging questions.

Finally, don’t underestimate the possibility of diversifying: monitoring Wikipedia, focusing on quality videos, or animating a community can help strengthen your authority. Overviews often cite sources such as Wikipedia or YouTube: if you can also bring value to these platforms, you gain a fast track to visibility, even in more competitive contexts.

AI Overview vs AI Mode: don’t get confused

To complicate matters further, there is another issue: as already announced in May at Google I/O 2025, AI Overview is not the only AI-based solution ready for Search, and Google is increasingly pushing the concept of “AI-first search.”

If we have familiarized ourselves with AI Overview and are coming to terms with it, the release of AI Mode in Europe seems increasingly close—in fact, the feature arrived on the Old Continent a few days ago (at least geographically), given its release in the United Kingdom. And the differences between the two products are enormous.

AI Overview does not change the overall structure of the page, but is inserted between the organic results and, increasingly, bypasses them thanks to its privileged position. AI Mode, on the other hand, represents a paradigm shift: it is an opt-in mode, which can only be activated by the user in certain countries (currently the United States, India, and, very recently, the United Kingdom), which transforms Google into a true conversational assistant. In this mode, the search is not limited to returning a list of links or a generated answer: it becomes a continuous dialogue between the user and artificial intelligence, with the ability to ask follow-up questions, upload documents and images, and progressively refine the results. This is the direction Google is heading in for the future, but at the moment it is only accessible to those participating in tests in selected markets.

AI Mode is not a “box,” but an alternative, immersive search mode that transforms Google into a real chatbot, similar to ChatGPT. It is activated (for now) voluntarily and allows you to:

• Ask complex, multi-step questions.
• Have a continuous dialogue with the AI to refine your search.
• Use advanced features such as “Deep Search” for in-depth analysis.
• Ask the AI about uploaded documents, such as PDFs and, from the end of July 2025, images too. Its rollout is slower and more controlled, mainly active in the US and other selected markets for Labs program subscribers, but it represents Google’s ultimate vision for the future of search.

A new beginning for SEO

AI Overview is not a passing fad or a simple feature. It is a fundamental change in the infrastructure of organic search, with a profound, measurable impact that is already underway. The Pew Research Center’s data does not lie: traffic to websites is declining for a wide range of queries, forcing the entire industry to rethink its strategies.

Understanding its multifaceted nature, learning to measure its effects with a data-driven approach, and starting to prepare for AI Mode are the first essential steps.

The challenge for all of us is no longer just “how to rank first to get the click,” but a much more complex one: “How do you become such an authoritative, reliable, and well-structured source that you are chosen by Google’s artificial intelligence as part of the answer?” It’s a new beginning for SEO, and those who are first to interpret it will have an invaluable competitive advantage.