54%
of Italian SERPs now display an AI Overview. In January 2025, it was 6.5%. In ten months, its presence has nearly decupled, with the European roll-out completed in October 2025.
AI Overviews and generative search: how the Italian SERP has changed and what a website is used for today.
Between January 2025 and May 2026, Google Search underwent deep changes. AI Overviews now appear in 54% of Italian SERPs, organic click-through rate has plummeted by 61% on queries where they show up, and the website — which until yesterday was the sole traffic destination — is turning into something else: a source from which language models extract answers. Everything changes: what is measured, how content is written, and the role of the webpage.
In summary. The introduction of AI Overviews has transformed search from an index of links to a synthesized response. In Italy, they appear in 54% of SERPs (SEOZoom, December 2025) and cause organic CTR to drop by 61% on informational queries (Seer Interactive, September 2025). The most impacted sectors are health, finance, education, and B2B tech, with coverage of up to 40% of queries. The website does not disappear: it becomes a source to cite, a structured archive, and proof of authority for language models. Traffic coming from AI is 4.4 times more qualified (Semrush). The reference KPI changes: no longer total clicks, but share of voice in AI responses.
Five data points describing the new SERP.
These are not projections: they are measurements taken between mid-2025 and early 2026 on millions of real queries.
−61%
is the drop in organic CTR on queries where an AI Overview appears. From 1.76% to 0.61%. The study analyzed 3,119 informational queries across 42 organizations, totaling 25 million impressions.
83%
of searches with an AI Overview present end without any click to a website. Out of total Google searches (with or without AIO), the zero-click rate is 60%. In Google AI Mode, it reaches 93%.
17–38%
is the share of top-10 pages that also appear as a source in AI Overviews. In mid-2025, it was 75%. Ranking first no longer guarantees being cited by the model.
4,4×
is the conversion multiplier for traffic arriving from AI platforms compared to standard organic traffic. The user arrives already informed, at an advanced decision-making stage. Ahrefs reports up to 23x in an internal case study.
−50%
is the estimated loss of organic traffic to websites by 2028 due to the expansion of generative search. This is not a catastrophist prediction: it is the trajectory implicit in current trends.
The SERP is no longer an index. It is an answer.
Until 2024, the Google results page was an ordered list of links. The algorithm decided who stood at the top, but the work of synthesis was always done by the user: they opened three or four results, compared them, and decided. The website was a necessary endpoint.
With the introduction of the Search Generative Experience (later renamed AI Overviews) first in the United States in 2024 and the completion of the European roll-out in October 2025, that mechanism broke. Today, on the majority of Italian informational queries, the user finds a block of text generated by Gemini at the top of the page summarizing answers extracted from 5–8 different sources. Below, traditional organic links continue to exist, but attention and clicks focus on the synthesized response.
On October 8, 2025, Google completed the release of AI Mode in Europe, the conversational search interface that operates like a dialogue rather than a dry query. In these interactions, the click-through rate to websites drops to 7%. The user no longer searches for "best CRM for SMEs"; they ask Google "what is the best CRM for my 20-person B2B services company?" and get a ready-made comparative recommendation.
In parallel, over 79% of consumers report regularly using AI search (ChatGPT, Perplexity, Copilot, Gemini) and 70% state they trust its answers. The website continues to exist, but the user encounters it in a new way: not as the destination of a query, but as the source of an answer generated elsewhere.
What has concretely changed.
| Traditional SERP (pre-2024) | SERP with AI Overview (2025–2026) | |
|---|---|---|
| Response format | List of 10 blue links | Generated synthesized response + list below |
| User effort | Open multiple sites, compare, synthesize | Read the already synthesized response, optionally dig deeper |
| CTR first result | ~28–30% | ~9–12% (average drop of 34–79% according to the study) |
| Zero-click rate | ~50% of total searches | 83% on queries with AIO, 60% in total |
| Ranking unit | Keyword-optimized page | Extractable short answer + authoritative entity |
| Predictability of classic ranking | High: top-10 = visibility | Low: top-10 overlap vs AI citations dropped from 75% to 17–38% |
| Volatility of cited sources | Relatively stable ranking | 70% of cited pages change every 2–3 months |
| Quality of residual traffic | Heterogeneous: from exploration to conversion | On average higher: 4.4× higher converting (Semrush) |
| Primary metric | Position + clicks | Share of voice in AI citations + traffic quality |
Which sectors in Italy are already undergoing the transformation.
The impact is not uniform. AI Overviews follow a logic of surgical replacement in areas where incorrect information can cause harm (Your Money Your Life), and growing coverage in sectors with a strong informational component.
Health
40.3% of AI Overview coverage on queries
It is the most impacted sector in Italy. Google replaces the publisher to ensure pre-validated medical answers. Generalist pages are losing ground; vertical brands with clear clinical authority, recognizable authors, and cited primary sources win.
Source: SEOZoom, December 2025
Finance
20.2% of AI Overview coverage
Second most covered YMYL sector. Comparative queries ("difference between mutual fund and ETF", "best savings account") are perfect fuel for LLMs, which synthesize scattered data. Generalist financial sites are losing traffic; regulated specialists gain citations.
Source: SEOZoom, December 2025
B2B Tech
~70% of AI Overview exposure
Buying committee queries ("best CRM for SMEs", "alternatives to Salesforce") are highly informational: AI Overviews cover them en masse. This is the segment where GEO has the highest ROI, because the customer arrives at the site already "screened" by the model.
Source: Digital Applied, 2026
Education & Consulting
+361% of AI Overview presence YoY
The fastest-growing sector for AI coverage in Italy. Courses, professional training, technical consulting: all structured informational queries, perfect for LLM synthesis. Schools/agencies with structured educational content win; those selling only the brand lose.
Source: BrightEdge, 2026
Publishing & Food
AI-ready high according to Madcrumbs
In Italy, according to the Madcrumbs AI Search Impact Index (Sep 2025 – Feb 2026), publishing and food are among the most "AI-ready" sectors: their structured and advisory content is already systematically cited by ChatGPT, Perplexity, and Copilot.
Source: Madcrumbs AI Search Impact Index, March 2026
E-commerce & Retail
~10% of commercial queries with AIO
The least impacted sector so far. When the user seeks to buy, Google knows a click is still needed. But beware: in categories with a strong comparative component (headphones -23 points of click share, jeans -17, greeting cards -13), erosion is already underway.
Source: ALM Corp, February 2026
Four new jobs for the web page.
The website does not die: it ceases to be primarily a destination. It becomes something else. Four new, complementary roles that coexist with the traditional one but invert its priority.
1. From destination to source
The site is no longer (only) where the user arrives: it is where the LLM crawler arrives to extract the answer to be shown elsewhere. A page that converts 100 users per month but is cited by ChatGPT on 5,000 relevant queries has a brand value unseen with traditional KPIs. The site works even when it receives no clicks.
2. From shelf to structured archive
A page without schema markup, structured data, an explicit author, or update dates is today invisible to models, even if well-positioned on the search engine. Technical structure (Schema.org, llms.txt, marked FAQs, connected entities) becomes more important than graphics and emotional copy.
3. From catalog to corpus of proof
The LLM does not trust marketing: it trusts verifiable facts. A product page that says "the best in Italy" will not be cited; a page that says "operating since 2009, over 200 clients, certification X, verified reviews Y" can become the cited source. The website serves to provide the model with sufficient proof to include you in the response.
4. From display to qualified conversion point
The (few) users arriving from AI are different: they already know what they want and have already compared. They convert 4.4 times more (Semrush). The site no longer needs to entertain and convince: it must close. The funnel changes: shorter forms, more factual product pages, more direct CTAs, highlighted social proof.
What companies must do now.
It is not the time to dismantle classic SEO: content that currently ranks in the top 10 is still the most likely to be cited by AI (even if the correlation has weakened). But a new layer must be added.
Restructure key content
The first 40–60 words of every important page must answer the primary query in a self-contained, model-citeable way. No preamboles, no storytelling: just the answer. Then the article can begin.
- Answer-first on top 50 content pages
- Marked FAQs on all commercial pages
- Verifiable data every 150–200 words
Implement the semantic skeleton
Complete schema markup, llms.txt, robots.txt open to AI bots, company knowledge graph with recognizable authors. This is the most technical work but with the most immediate ROI: language models treat the site as a reliable and extractable source.
- Schema Organization + Person + Service
- Complete FAQ + HowTo + Article
- llms.txt in the root
- Authors' knowledge graph
Build measurable authority
Models reward vertical specialists over generalists. Better to dominate a thematic niche with 30 deep content pages than produce 300 articles on everything. Brand the author, build a professional profile, use real reviews, and make certifications explicit.
- Author page with bio + sameAs
- Citations of primary sources
- Thematic verticalization
- Verifiable reviews and case studies
What to measure today (and what to stop measuring).
Traditional metrics (position, CTR, sessions) are not dead but have become insufficient. They must be accompanied by metrics that measure presence in AI answers, not just traffic to the site.
| Old metric (still useful) | New metric (now priority) | Tool |
|---|---|---|
| Average SERP position | Share of voice in AI: how many times you are cited on relevant prompts | Otterly, Profound, BrandRank.AI |
| Organic CTR | Citation overlap: % of top-10 pages also cited in AI Overviews | Manual + AI Overview Monitor |
| Total sessions | AI-referred traffic: visits from ChatGPT, Perplexity, Copilot, Gemini | GA4 + source filter (chat.openai.com, perplexity.ai, etc.) |
| Pageviews | AI traffic quality: duration, scroll depth, conversions | GA4 + custom segments |
| Bounce rate | AI-referred conversion rate vs other sources | GA4 + attribution |
| Quantitative backlinks | Brand mentions + recognized entity in the knowledge graph | Google Knowledge Graph Search API + brand monitor |
| Organic traffic volume | AI-assisted pipeline: leads declaring they found you via ChatGPT/Perplexity | Form — "how did you find us" field |
The right question is not "does the website still serve a purpose?". It is "what purpose does it serve today?".
The narrative of recent weeks oscillates between two extremes: some announce the death of the traditional web, while others minimize it ("it's just another Google feature, like featured snippets ten years ago"). Both positions miss the point.
The concrete fact is this: the website continues to exist, receives fewer clicks, and the value of those few remaining clicks is significantly higher. At the same time, the website acquires a new utility unseen before: it feeds AI responses displayed elsewhere. A brand cited by ChatGPT on a relevant query is more visible than before, even if it receives no visit.
For many sectors in Italy — health, finance, B2B tech, education, consulting, food — the window of adaptation is measured in months, not years. AI Overview coverage went from 6.5% to 54% of Italian SERPs in ten months. Companies that have already restructured their key content in an answer-first format, implemented complete schema markup, and built a vertical authority profile are beginning to be systematically cited. Others are losing visibility without realizing it, because their KPIs continue to "look" normal.
There is no magic formula. There is a change of discipline: measuring new things, writing differently, and giving the site the technical structure to be read by a machine even before a person. SEO does not die; it transforms. The website does not die; it changes jobs. Companies that understand this now, while 80% of the Italian market is still trying to return to the "old normal," build a competitive advantage that will be hard to catch up with.
On AI search and the new role of the website.
What are Google's AI Overviews?
AI Overviews are responses automatically generated by Google's Gemini model that appear above traditional search results, summarizing information from various web sources. In Italy, they appeared in 54% of SERPs at the end of 2025 (SEOZoom) and 48% globally (BrightEdge). Their main effect is to reduce clicks to websites because the user finds the answer directly on the search results page.
How much has organic traffic dropped with AI Overviews?
The Seer Interactive study from September 2025 documents a 61% drop in organic CTR on queries with AI Overviews (from 1.76% to 0.61%). Ahrefs, across 300,000 searches, measures a 34.5% drop in the CTR of the first position. For informational queries, the drop in traffic reaches 30–40%. Zero-click searches have reached 60% of total Google searches, and 83% on queries where an AI Overview is present.
Which Italian sectors are most impacted by AI Overviews?
According to the SEOZoom study of December 2025, in Italy the Health sector is the most impacted with 40.3% AI Overview coverage, followed by Finance (20.2%) and Family & Community (19.9%). These are all YMYL (Your Money Your Life) sectors where Google replaces the publisher to ensure verified answers. Transactional sectors like e-commerce and retail are less affected: only 10% of commercial keywords trigger an AI Overview.
Is a website still useful if clicks are collapsing?
Yes, but with a different function. The website stops being primarily a traffic destination and becomes the source from which AI models extract information to generate answers. Companies cited in AI Overviews get 35% more organic clicks and 91% more paid clicks compared to those not cited (Seer Interactive). Additionally, traffic coming from AI is 4.4 times more qualified in terms of conversions (Semrush) because the user is already informed and in the decision-making stage.
Which metrics should we start measuring today?
Traditional metrics (SERP rankings, CTR, total sessions) are no longer sufficient. They must be accompanied by share of voice in AI responses (one's own share of citations compared to competitors on a set of relevant prompts), AI-referred traffic (visits from ChatGPT, Perplexity, Copilot, Gemini), citation overlap rate (how many of one's ranking pages also appear as sources in AI Overviews), and the quality of AI-referred traffic (time on page, conversions, average value). Tools: Otterly, Profound, BrandRank.AI to monitor citations; GA4 for AI-referred traffic.
What should Italian companies do now?
Three concrete actions: (1) restructure key content into an answer-first format, with self-contained answers extractable by models (40–60 verifiable words); (2) implement complete schema markup (Organization, Article, FAQPage, Service, Product) to give models a semantic skeleton of the site; (3) build real thematic authority through specialized content, citations of primary sources, a structured author profile, and verified reviews. The first share of voice results are measurable in 4–8 weeks; building full authority takes 3–6 months.
Is traditional SEO still useful or is it obsolete?
Traditional SEO remains the foundation: AI Overviews tend to cite content already well-positioned in SERPs, even though the correlation has weakened (top-10 overlap dropped from 75% to 17–38%). Backlinks remain a primary ranking factor. What changes is that ranking on the first page is no longer enough: you must also be structured for fact extraction by the models. GEO (Generative Engine Optimization) works alongside SEO, not as a replacement.
Is your company visible in AI responses?
Free audit of your presence in AI search (ChatGPT, Perplexity, Google AI Overview, Copilot). We measure where you are already cited, where your competitors appear instead of you, and the concrete actions to take in the next 90 days.
Request free GEO audit
Document type: market case study — industry analysis based on verifiable public data.
Published: May 15, 2026 — Last updated: May 15, 2026
Author: Di Marco Nardino, Startupweb.io
Cited sources: SEOZoom (Dec 2025), Seer Interactive (Sep 2025), Ahrefs, Semrush, BrightEdge, Madcrumbs AI Search Impact Index (Mar 2026), Digital Applied, ALM Corp (Feb 2026), SparkToro, Gartner.
Related resources.
Service
GEO & AI Search
The operational service: how we make you citeable by ChatGPT, Perplexity, AI Overviews, and Copilot.
Discover the service ServiceSEO Consulting
Traditional SEO remains the foundation: without an organic top 10, AI citations are much harder.
Discover the service IndexAll case studies
Return to the complete list of industry analyses and client case studies of Startupweb.io.
See all