- What Google has demonstrated and why you should care
- Should you worry now or is it just theory?
- What is already happening to the organic traffic of Italian businesses?
- What changes concretely for those who manage a corporate website?
- Three concrete scenarios: how much does this transition matter for an Italian company?
- What tools and approaches exist to adapt to this new logic?
- What rules and risks do you need to keep under control?
- Why people are talking about it now: has the market really changed?
- How to get started concretely: the steps to adapt the strategy?
- What are the most common mistakes to avoid at this stage?
- How to tell if your strategy is working?
- So, is it worth changing your SEO strategy right away for your business?
- How we see it at SHM
- Frequently Asked Questions
- Google has proven on paper that an AI model can do the job of a search index: no list of results, just a generated answer.
- Visibility becomes binary: either you are in the set of content the model produces, or you don't exist — there's no longer a 'page 2' to recover from.
- Organic CTR in position 1 has already plummeted by 58% when an AI Overview appears (Ahrefs, December 2025): the change is happening now, not in the future.
- The right strategy is no longer 'climbing the rankings' but 'being chosen as a source' by Google's AI systems.
Google has published mathematical proof that an AI model can rank an unlimited number of documents without consulting a separate index. If this architecture becomes reality, the search results page — the one you know today — might simply cease to exist.
What Google has demonstrated and why you should care
In January 2026 Google published research that mathematically proves how an AI model can classify content without a separate index. It is not a product yet, but it confirms a direction Google has been working on publicly since 2021.
Today Google works like this: a crawler (an automatic program) visits billions of web pages, archives them in a index — a sort of giant catalog — and then, when you search for something, it retrieves the most relevant results from that catalog. Google's index exceeds 100 million gigabytes and contains hundreds of billions of pages, according to AllOutSEO .
The research published in January 2026 shows that a language model—the same type of AI behind ChatGPT—can do this job without a separate catalog. The model learns the content during training and generates directly when answering a question. No list of links: just an answer.
For your business, this isn't just a technical update to brush off. It's a sign that traditional SEO — tweaking pages to climb up a list — might stop making sense in just a few years.
| How it works today | How it might work tomorrow |
|---|---|
| Google indexes pages into a catalog | The AI model learns the content during training |
| Search retrieves results from the catalog | The model directly generates the answer |
| See a list of links (SERP) | You see an answer, without a list of results |
| 'Indexed' = visible in search | ‘Known by the model’ = visible in the answer |
| Page 2 exists but gets few clicks | Nothing exists beneath the generated set |
Should you worry now or is it just theory?
It's theory, but backed by five years of public research and with impacts already showing on organic traffic. The shift is happening step by step, but today's numbers prove the trend is well underway.
The January 2026 research is not a product announcement. The experiments were done on small datasets—82,000 concepts and 310 shopping examples—and the model used is not the one Google uses in production. The original article itself on Search Engine Journal defines it as «a foundation, not a deployment».
But Google doesn't publish five years of research in the same direction out of academic curiosity. The path is clear: proposal in 2021, prototype in 2022, mathematical proof in 2026. Every step has brought this architecture closer to production.
Meanwhile, the effects are already visible on traffic data. Searches that end without any clicks to external sites have jumped from 56% to 69% in just one year, according to Similarweb . It is not a future problem: it is a problem of today.
In Italy the situation is slightly better compared to the US, but the trend is the same. According to SparkToro and Similarweb, in early 2026 63.4% of Italian searches on Google end without a click to external sites, compared to 68% in the US in the same period, as reported by Clickforest .
What is already happening to the organic traffic of Italian businesses?
Organic traffic is shrinking visibly. Sites that aren't mentioned as a source in Google's AI Overviews are already getting less than half the clicks compared to queries without an AI Overview.
The AI Overview are the generated answer boxes that Google shows at the top of search results, before any links. They arrived in Italy in February 2025. The AI Mode — the complete conversational version, without a traditional results list — was rolled out to European traffic in October 2025, according to rafaelpatron.com.
Data measured by Ahrefs in December 2025 shows that the CTR (click-through rate, meaning the percentage of people who click on a result) in position 1 drops by 58% when an AI Overview is present, compared to a 34.5% drop measured in April 2025. Almost double in nine months, as reported by Search Engine Journal .
The difference between being cited and not being cited is huge: 2.1% versus 0.9% CTR, according to the Seer Interactive study on over 5 million queries, reported by SpazioRC . It is not a matter of ranking position: it is a matter of inclusion or exclusion.
The Pew Research Center measured that when an AI Overview appears only 8% of users click on a traditional organic result, compared to 15% when the AI Overview is not present. Only 1% click on the links cited inside the AI Overview itself, as reported by Myoho Marketing .
What changes concretely for those who manage a corporate website?
The logic shifts from 'climbing rankings' to 'being chosen as a source.' It is no longer about optimizing for a leaderboard, but about becoming a recognizable reference for Google's AI systems.
In Google search as of January 2026, every document is no longer just a URL but a docID : a short code that the model generates directly, as if it were writing a word. Two very similar pages don't compete for adjacent positions: they compete for the same code, and one of them loses out entirely.
This has three practical consequences for those who manage online content.
- 1 · Content It has to stand out Similar pages don't compete for positions: they cancel each other out. Generic or duplicate content loses visibility permanently.
- 2 · Quality signals They become the algorithm The model learns from clicks, human ratings, and engagement signals. There is no separate layer of 'ranking factors' to optimize: there is a learned preference distribution.
- 3 · Updates They take different amounts of time In a traditional index, a new page is crawled and indexed. In a generative model, it has to be 'learned'. Publishing no longer means being known by the system.
- 4 · Visibility Becomes binary Either you are in the model's generated content set, or you don't exist. There is no page 2 to recover from: nothing is produced below the generated set.
You can learn more about how this transition is already reshaping the rules of search in the article on Google Search becoming a universal AI assistant , which analyzes the evolution of the search interface toward a conversational model.
Three concrete scenarios: how much does this transition matter for an Italian company?
The impact depends on how much your business relies on organic traffic and how unique your content is compared to the competition.
These scenarios are illustrative: the numbers serve to show the order of magnitude, they are not real client data.
| Company type | Current situation | Concrete risk | Priority action |
|---|---|---|---|
| E-commerce with a large catalog | Many similar product pages | High: duplicate pages lose visibility altogether | Differentiate product pages with unique and specific content |
| Professional practice or consulting | Blogs with generic articles on common topics | Medium-high: content replaceable by the AI Overview | Focus on real cases, proprietary data, original perspective |
| B2B manufacturing company | Tech site with low content output | Medium: less competition, but also fewer signals for the model | Build deep technical content linked by authoritative sources |
| Local store network | Reliance on local searches and Google Maps | Low in the short term: local searches change more slowly | Manage Google Business Profile and local citations |
What tools and approaches exist to adapt to this new logic?
Platforms and methodologies already exist to measure visibility in AI systems, not just in traditional SERPs. The choice depends on your company's digital maturity.
| Approach | What it measures or does | Who it is for | Main limitation |
|---|---|---|---|
| Traditional SEO (rank tracking) | Keyword ranking positions | Who still has significant organic traffic | It doesn't measure visibility in AI systems |
| AI Visibility monitoring | Citations and inclusion in AI answers (Google, Perplexity, ChatGPT) | Companies with medium-high budgets and dependence on informational search | Market still young, non-standardized metrics |
| Content authority building | Builds topical authority with deep and original content | Any company with editorial resources | Takes time: not a quick fix |
| Structured data and schema markup | Help AI models understand and cite your content | Sites with catalogs, FAQs, articles, events | Requires technical expertise for implementation |
| Multi-channel presence (social, newsletter, PR) | Reduces dependence on Google organic traffic | Companies with an existing or buildable audience | Does not replace search visibility |
The SEO and visibility for AI is a topic we are following closely: you can find all the articles analyzing this transition in the cluster hub, from AI Overviews to AI Mode.
It is also worth reading the analysis on Google paying websites for AI answers : a sign that Google is also trying to strike a balance between the new generative model and the survival of the publishing ecosystem.
What rules and risks do you need to keep under control?
The main risk is not technical but strategic: continuing to invest in metrics that measure a world that is changing, without realizing that the numbers no longer tell the real story.
Google's research describes a system where position 11 is not a poor result: is never generated . The model produces a set of results through a process called beam search (beam search): generates the most likely sequences and discards everything else. If you are not in the beam, you do not exist.
This creates a measurement problem. Today you can track your position for thousands of keywords. Tomorrow, the right question won't be "where do we rank?" but "are we in the generated set, for which queries, and how often?" These are two different questions, and they require different tools.
A second risk concerns duplicate or generic content. If you publish very similar pages — almost identical product sheets, articles covering the same topic with minimal variations — you are already losing visibility. In generative architecture, this problem is aggravated: indistinguishable pages compete for the same space, and one of them is eliminated entirely.
| Risk | Warning sign | what to do |
|---|---|---|
| Measuring only ranking positions | Traffic drops but rankings seem stable | Add AI visibility metrics and real CTR |
| Duplicate or generic content | Many similar pages with low or zero traffic | Consolidate, differentiate, or delete weak pages |
| Dependency on a single traffic source | Over 70% of traffic comes from Google organic search | Diversify: newsletter, social, PR, direct search |
| Ignoring engagement signals | High bounce rate, low time on site | Improve content quality and relevance |
| Not updating existing content | Articles and pages untouched for over 12 months | Periodic review: the AI model rewards freshness |
Why people are talking about it now: has the market really changed?
Yes. Data from 2025 and 2026 shows that the transition is already happening, it is not a future prediction. Companies that wait to see the impact on their traffic before taking action risk starting too late.
Google processes over 9.5 million searches per minute, for a total of 5 trillion searches per year, according to AllOutSEO . Even a variation of a few percentage points in user behavior translates into millions fewer clicks for websites.
The drop in organic CTR from -34.5% in April 2025 to -58% in December 2025 — measured by Ahrefs on position 1 in the presence of AI Overviews — shows a speed of change that leaves little time to adapt calmly.
The article on Google self-expanding AI Overviews analyzes how this automatic expansion is already squeezing organic traffic on increasingly broad queries. And the analysis on impact of AI Overviews on SEO in 2026 offers a complete picture of the strategies available for Italian SMEs.
How to get started concretely: the steps to adapt the strategy?
No need to wait for Google's generative architecture to be in production. The useful actions to prepare are the same ones that improve AI visibility today.
- 1 · First step Measure where you really stand Check the actual CTR of your top pages in Google Search Console, not just your rankings. If your CTR is under 1% on informational queries, an AI Overview is probably already stealing your spotlight.
- 2 · Second step Identify content at risk List the pages with dropping traffic over the last 6 months. Check if an AI Overview appears for those queries. These are the contents to rewrite or consolidate with priority.
- 3 · Third step Make every piece of content unique Each page must have something that cannot be found anywhere else: proprietary data, real-world cases, a specific perspective, direct experience. Generic content is replaced by the AI Overview, while distinctive content gets cited.
- 4 · Fourth step Add structured data Structured data (semantic markup) helps AI models understand what your page is about and cite it correctly. FAQs, articles, products, and events have specific schemas to implement.
- 5 · Fifth step Monitor AI visibility Start tracking if and when your site is mentioned in Google AI Overview, Perplexity, or ChatGPT responses. There are dedicated tools, still young but already useful for understanding the direction.
- 6 · Sixth step Diversify traffic sources Newsletters, social media, PR, and direct search reduce dependence on Google. This isn't giving up on SEO: it's a safety net while the rules change.
What are the most common mistakes to avoid at this stage?
The most common mistake is continuing to optimize for yesterday's metrics while the system shifts beneath your feet. The second is waiting for the change to be complete before making a move.
| Error | Why it is a problem | Correct alternative |
|---|---|---|
| Publishing generic content for volume | They are replaced by the AI Overview without being cited | Fewer, deeper, and more distinctive contents |
| Measuring only ranking positions | Rankings no longer tell the story of real traffic | Monitor CTR, impressions, and AI citations |
| Ignoring pages with declining traffic | The problem gets worse with every expansion of AI Overviews | Quarterly audit and review of weak content |
| Treating all queries the same way | Informational queries are already dominated by AI; transactional ones less so | Segment the strategy by query type |
| Wait until generative architecture is in production | The change is already underway: every month of delay costs traffic | Act on the basics (quality, structure, diversification) right away |
How to tell if your strategy is working?
The metrics to watch are changing: no longer just rankings, but real CTR, citations in AI systems, and traffic diversification.
| Indicator | What it measures | Positive signal | Tool |
|---|---|---|---|
| Organic CTR per page | How many people click compared to those who see the result | Stable or growing despite AI Overviews | Google Search Console |
| Impressions on informational queries | How many times the site appears in answer searches | Growing or stable | Google Search Console |
| Mentions in AI Overviews | If and when the site is cited as a source | Citations present on relevant queries | AI visibility tools (e.g., Semrush, Ahrefs) |
| Traffic from sources other than Google | Reliance on a single source | Google share below 60% of total | Google Analytics / GA4 |
| Time on site and pages per session | Quality of incoming traffic | Growing: signal of relevant content | Google Analytics / GA4 |
| Backlinks from authoritative sources | Authority perceived by AI models | Steady growth of links from industry sites | Ahrefs, Semrush |
So, is it worth changing your SEO strategy right away for your business?
It is worth it if over 50% of your traffic depends on Google organic search, you have generic or duplicate content, and you operate in sectors where informational queries are frequent (consulting, training, e-commerce of common products). Every month of delay translates into lost traffic.
A middle ground you already have good traffic diversification and quality content, but you are not yet monitoring visibility within AI systems. You can take a gradual approach, but start measuring real CTR and AI citations right away.
Not yet if your business depends almost exclusively on local searches, direct relationships, or offline channels. The impact is slower, but it is still worth starting to build authoritative content for the medium term.
How we see it at SHM
For us, the January 2026 Google search isn't just tech news for insiders: it's the clearest sign yet that SEO as a race for rankings is ending. And that the real competition is moving to a different ground: be recognizable and authoritative in the eyes of AI models , not just in the eyes of ranking algorithms.
We believe AI should be integrated to free companies from routine and give back time, not to chase every new trend as if it were a fad. This also applies to SEO: it's not about adopting new tools in a rush, but understanding what really changes in visibility logic and methodically adapting the strategy.
With some clients we're already working on this: content audits to spot at-risk pages, structured data to make things easier for AI models to read, and tracking mentions in generative response systems. It's not something you do in a week, but it's what makes a difference in the medium term.
If you want to understand where your site stands regarding this transition, our services of SEO for AI and of strategic AI integration they are the starting point. For those who want to build a solid digital presence in the long run, we also work on site structure and architecture and on measurement of visibility data .
Frequently Asked Questions
Not right away: the January 2026 Google search is a proof of concept, not a live product. In the short term, the most likely setup is hybrid: a classic index powering a generative model for the final answer. But the direction is clear, and those who don't prepare risk losing visibility step by step.
It's still needed, but it's no longer enough. The basics — quality content, solid technical structure, authoritative links — remain valid because they're the same signals AI models use to assess authority. What changes is the goal: no longer climbing positions in a list, but being chosen as a source by generated answers.
Google Search Console shows impressions and clicks, but it doesn't separate mentions inside AI Overviews from traditional organic results. Tools like Semrush and Ahrefs are rolling out specific features to monitor AI visibility. For more precise tracking, there are dedicated platforms out there that are still evolving.
No, but you need to figure out which content is at risk. The starting point is an audit: which pages are seeing a drop in traffic? Is an AI Overview showing up for those queries? Generic or duplicate content needs to be combined or rewritten with a unique angle. Content that is already distinct and authoritative usually doesn't need immediate changes.
It depends on where you start. A site with good authority and quality content can see improvements in AI citations in 3-6 months. A site with weak content or problematic technical structure requires foundational work first. In any case, it's a medium-term strategy: there are no shortcuts.
Paid campaigns are not directly affected by the transition to generative search, at least in the short term. Google has an interest in maintaining advertising as a source of revenue. But if organic traffic decreases, the relative weight of paid campaigns increases, and with it the costs. A balanced strategy between organic and paid becomes even more important.
The question is no longer "where does my site appear?" but "is my site chosen as a source when AI answers instead of me?"
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
