SEO without Google's index: what changes for your site's visibility

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 you know today could simply cease to exist.

What has Google shown, and why should you care?

In January 2026, Google published research mathematically proving that an AI model can rank content without a separate index. It's not a product yet, but it confirms a direction Google has been working on publicly since 2021.

Here's how Google works today: a crawler (an automated program) visits billions of web pages, stores them in an index, a sort of gigantic catalog, and then, when you search for something, 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 it directly when answering a question. No list of links: just an answer.

For your company, this isn't a technical update to ignore. It's a sign that classic SEO, optimizing pages to climb positions in a list, could lose its meaning within a few years.

Should you worry now, or is it just theory?

It's theory, but with five years of public research behind it and consequences already measurable in organic traffic. The change is gradual, but today's numbers show the direction is already underway.

The January 2026 research is not a product announcement. The experiments were run 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 on Search Engine Journal itself calls it "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. Each step has moved this architecture closer to production.

Meanwhile, the effects are already showing in traffic data. Searches that end with no click to external sites rose from 56% to 69% in a single year, according to Similarweb. It's not a future problem: it's a problem today.

In Italy the situation is slightly better than in the US, but the trend is the same. According to SparkToro and Similarweb, in early 2026 63.4% of Italian Google searches end with no click to external sites, compared with 68% in the US over the same period, as reported by Clickforest.

What's already happening to organic traffic for Italian companies?

Organic traffic is shrinking in a measurable way. If you're not cited as a source in Google's AI Overviews, you already get less than half the clicks of a site on a query with no AI Overview.

AI Overviews are the generated answer boxes Google shows at the top of search results, before any link. They arrived in Italy in February 2025. AI Mode, the full conversational version with no traditional list of results, was extended to European traffic in October 2025, according to rafaelpatron.com.

Data measured by Ahrefs for December 2025 shows that CTR (click-through rate, the percentage of people who click a result) in position 1 drops by 58% when an AI Overview is present, compared with a 34.5% drop measured in April 2025. That's almost double in nine months, as reported by Search Engine Journal.

The difference between being cited and not being cited is huge: a 2.1% CTR versus 0.9%, according to the Seer Interactive study of more than 5 million queries, as reported by SpazioRC. It's not about ranking position. It's about being included or excluded.

Pew Research Center found that when an AI Overview appears, only 8% of users click a traditional organic result, compared with 15% when no AI Overview is shown. Just 1% click the links cited inside the AI Overview itself, as reported by Myoho Marketing.

What concretely changes for those who run a company website?

The logic shifts from "climbing the rankings" to "being chosen as a source." It's no longer about optimizing for a ranking, but about becoming a recognizable reference for Google's AI systems.

In Google's January 2026 research, every document is no longer a URL but a docID: a short code 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 anyone managing content online.

You can find out more about how this transition is already redrawing the rules of search in our article on Google Search becoming a universal AI assistant, which looks at how the search interface is evolving toward a conversational model.

Three real-world 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 differentiated your content is from your competitors'.

These scenarios are illustrative: the numbers show the order of magnitude and are not data from real clients.

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 right choice depends on your company's digital maturity.

SEO and AI visibility is a topic we're following closely: in the cluster hub you'll find all the articles that analyze this transition, from AI Overviews to AI Mode.

It's also worth reading our analysis of Google paying websites for AI answers: a sign that Google, too, is looking for 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 isn't technical but strategic: continuing to invest in metrics that measure a world that's changing, without noticing that the numbers no longer reflect reality.

Google's research describes a system where position 11 isn't a poor result: it's never generated. The model produces a set of results through a process called beam search: it generates the most probable sequences and discards everything else. If you're not in the beam, you don't 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?" Those are two different questions, and they require different tools.

A second risk involves duplicate or generic content. If you publish pages that are very similar to each other, like nearly identical product pages or articles covering the same topic with minimal variations, you're already losing visibility. In a generative architecture, this problem gets worse: indistinguishable pages compete for the same space, and one of them is eliminated entirely.

Why is everyone talking about it now: has the market really changed?

Yes. The 2025 and 2026 data show the transition is already underway. It's not a future forecast. Companies that wait to see the impact on their own traffic before acting risk starting late.

Google handles over 9.5 million searches per minute, for a total of 5 trillion searches a year, according to AllOutSEO. Even a shift of a few percentage points in user behavior means millions fewer clicks for websites.

Organic CTR fell from -34.5% in April 2025 to -58% in December 2025, as measured by Ahrefs for position 1 when an AI Overview is present. That pace of change leaves little time to adapt at a relaxed pace.

Our article on Google auto-expanding AI Overviews looks at how this automatic expansion is already squeezing organic traffic on ever broader queries. And the analysis of the impact of AI Overviews on SEO in 2026 gives a full picture of the strategies available to Italian SMEs.

How do you get started: what are the steps to adapt your strategy?

You don't need to wait for Google's generative architecture to go live. The steps that prepare you are the same ones that improve your AI visibility today.

What are the most common mistakes to avoid at this stage?

The most common mistake is to keep optimizing for yesterday's metrics while the system shifts under your feet. The second is waiting for the change to be complete before making a move.

How can you 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.

So, is it worth changing your SEO strategy right now for your company?

It's worth it if more than 50% of your traffic depends on Google organic search, you have generic or duplicate content, and you operate in sectors where informational queries are common (consulting, training, e-commerce for everyday products). Every month of delay means lost traffic.

Somewhere in between: you already have well-diversified traffic and quality content, but you're not yet monitoring your visibility in AI systems. You can move gradually, but start measuring real CTR and AI citations right away.

Not yet if your business depends almost entirely on local searches, direct relationships or offline channels. The impact is slower, but it's still worth starting to build authoritative content for the medium term.

How we see it at SHM

To us, Google's January 2026 research isn't a technical story for insiders. It's the clearest sign yet that SEO as a race for rankings is coming to an end. And the real competition is moving to a different field: being recognizable and authoritative in the eyes of AI models, not just ranking algorithms.

We believe AI should be integrated to free companies from routine and give time back, not to chase every novelty as if it were a trend. That applies to SEO too: it's not about adopting new tools in a hurry, but about understanding what really changes in the logic of visibility and adapting your strategy with method.

We're already working on this with some clients: content audits to identify at-risk pages, structured data to make content easier for AI models to read, and monitoring of citations in generative answer systems. It's not a one-week job, but it's what makes the difference in the medium term.

If you want to understand where your site stands in this transition, our SEO for AI and strategic AI integration services are the place to start. If you want to build a solid digital presence for the long run, we also work on site structure and architecture and on measuring visibility data.

Frequently Asked Questions

The question is no longer “where does my site appear?” but “is my site chosen as a source when AI answers in my place?”