The fast fish eats the slow fish •

Perché nell'era dell'AI il tempo non è più solo denaro, ma una questione di sopravvivenza

For a hundred years, one rule alone held in business: the big fish eats the small fish.

And so revenue beat revenue, and the old dominated the new.
Every. Single. Time.

Yet today this rule is dead, and the one replacing it, which already governs winning decisions in boardrooms, is that the fast fish eats the slow fish. It does not matter how big it is. It does not matter how much experience it has. Speed in occupying a territory before others realise that territory exists is the key to everything.

Because, let’s be honest: the slow fish usually doesn’t realise it is slow. It only realises when it sees the others disappear from the frame.
This article, by contrast, is in a sense a snapshot of that terrain, and of this new ecosystem.

Today, 28 May 2026, four forces are at work, but they produce a single outcome. Those who decide in the next six months how to be found by artificial intelligence will build a competitive advantage that lasts for years. Those who don’t will be eaten.

In this manifesto, written by the Bliss Agency strategy team on the basis of the latest data from McKinsey, Forrester, Adobe, Bain & Company, Similarweb, Pew Research and the leading independent observatories on AI search, you will find:

The four fronts open right now (generative AI, Generative Engine Optimization, Google’s AI Overviews, AI governance), with the real figures for 2025-2026;
Because on these fronts every month of delay carries a compounding cost that keeps multiplying;
What the companies securing a place in the front row are doing (and what those being left behind are ceasing to do);
the regulatory deadline that few Italian entrepreneurs know about and that could become the biggest problem of autumn 2026;
What to do on Monday morning so as not to miss the boat.

You probably won’t find a reassuring tone in these pages. There is urgency here, because urgency is the fact. Anyone looking for an article that says everything will be fine and there is still time will no doubt find one somewhere online. What you will find here is the exact opposite.

Four converging forces, a single window of time

To understand why this is the precise moment when everything is decided, we need to bring into focus four forces that are all pushing at the same time in this phase. And it is precisely their simultaneity that makes this window unique.

Force one: generative AI has stopped being a technological curiosity and has become a channel of access to the market.

ChatGPT reached 800 million weekly active users in October 2025 (doubling in eight months from 400 million in February) and passed 900 million in February 2026, according to figures reported by TechCrunch and picked up by the leading global observers. Perplexity now processes 780 million queries a month, against 230 million in August 2024.
The question “Will people use AI to search for things?” has already been answered, and the answer is: “they are using it for almost everything”.

Force two: Google is dying, and fast.

Similarweb data reported by Search Engine Journal confirm that zero-click searches (those in which the user finds the answer on the results page and clicks on no link) rose from 56% to 69% between May 2024 and May 2025. Google’s AI Overviews appear in roughly 20-30% of queries (depending on the sector), and when they do, the CTR of the first organic result plummets by around 79%, according to Authoritas data. The Daily Mail has reported click-through drops of up to 89%. Business Insider saw its organic traffic fall by 55% between April 2022 and April 2025, forcing the company to cut 21% of its staff in May 2025 (source: Wall Street Journal). And every one of these figures represents revenue wiped out.

Force three: a new discipline is emerging, one that decides who gets cited by AI.

According to independent projections reported by Omnibound (2026), the GEO market will grow from 848 million dollars in 2025 to 33.7 billion in 2034, a CAGR of 50.5%. It is, in proportion, the same trajectory SEO followed between 2005 and 2015. And here is the figure that should set alarm bells ringing in Italian boardrooms: 47% of global brands still have no GEO strategy (Dataslayer, October 2025). 92% say they intend to implement one, but only 40.6% are actually doing so. The gap between intention and action is the exact definition of the competitive opportunity for those who move now.

Force four: European AI regulation is about to enter its enforcement phase.

The EU AI Act, Regulation (EU) 2024/1689, provides that from 2 August 2026 the strictest obligations for high-risk AI systems come into force (Articles 9-17 for providers, Article 26 for deployers). Even though in May 2026 the EU reached a political agreement to postpone some obligations (Travers Smith, May 2026), the direction of travel is unequivocal: companies using AI for decisions affecting people (recruitment, credit scoring, educational assessments, profiling) will need documented AI governance. Only those who arrive prepared will win. The unprepared will pay first with fines of up to 7% of annual global turnover, and then with the time lost playing catch-up.

Four forces converging on a six-to-twelve-month window. And only those who move now, within this window, will be able to withstand the pressure that follows in the years ahead.

Waiting until January 2027 will already be too late: the ground will be taken, AI citations will have consolidated around competitors, regulatory obligations will have to be handled as an emergency, and customers will already be used to no longer searching.

This is the scenario.

GEO: the new positioning

Ten years ago it would have seemed like science fiction, but today it is the norm: for the first time in the history of the internet, users see an answer directly, not a list.

That answer is generated by an artificial intelligence model that has chosen, from hundreds of possible sources, which ones to cite.

The break with the past is radical. In classic SEO, the goal was to reach first position for the keyword: the user saw ten links, the first took 30% of clicks and the tenth 1%. In Generative Engine Optimization, by contrast, the goal is to be one of the three, four or five brands the AI cites in its answer. And so, you are either in or you are out.

The criteria for this selection have changed. Research by Brandlight, cited by LLMrefs (March 2026), shows that the overlap between pages ranking first on Google and the sources cited by AI has fallen from 70% to below 20%. Put simply: ranking first on Google does not mean being cited by AI.
These are two different dynamics, with different rules. Neither rules out the other.

Data from Evertune Research, based on 200 million prompts analysed between October 2025 and February 2026, show that ChatGPT cites Wikipedia, YouTube, Reddit and LinkedIn as its most frequent sources, yet no single domain exceeds 5% of total citations. This means the “AI citation market” is more fragmented than the SEO one: there is room for a great many niche brands that, in traditional SEO, could never have competed.

And so big fish and small fish find themselves swimming in a new pond, where no one has a clear advantage any more. Not, at least, in the terms and logic of just a few years ago.

This is the first big piece of news for Italian SMEs: GEO is a democratic window of opportunity of a kind the internet has not offered in twenty years.
A mid-sized brand, well positioned and well structured, can appear in ChatGPT’s answers alongside multi-billion multinationals. How? By moving now, before AI systems consolidate their “preferences” around larger domains.

The second piece of news, however, is even more significant. The quality of AI traffic is far higher than that of traditional organic traffic. Data from Seer Interactive (June 2025) show that visitors from ChatGPT convert at 15.9%, from Perplexity at 10.5% and from Claude at 5%. The average conversion rate of Google organic traffic is 1.76%. This means that a visitor from AI is worth, on average, more than a traditional organic visitor. For those who move now, every euro invested in GEO is worth much, much more than the same euro spent on classic SEO.

AIO: you are Google’s first result, but nobody sees you any more

Google’s AI Overviews (the AI-generated summary that appears at the top of search results) are the most underestimated phenomenon of 2025. And that is because many business owners and marketing directors have simply not yet opened Google and seen what has changed.

Pew Research, drawing on 68,879 real searches tracked across 900 American adults in March 2025, established that 58% of users see at least one AI Overview in their everyday use of Google. Bain & Company (February 2025) found that 80% of consumers rely on AI Overview zero-click results for at least 40% of their searches. Adobe (July 2025) reports that 77% of American ChatGPT users use it as a search engine, and almost one in four already prefers it to Google.

Translated for a brand: the page you have invested three years of SEO in, which ranks first for a key keyword, is now clicked by less than half as many users as a year ago. And not because Google has demoted you. There is simply a summary above you now: and that summary is far easier to read.

You would do well to appear in it, if you do not want to be ignored.

This is the extinction of the click as a measure of value. The marketing metrics of the last twenty years (CTR, organic sessions, page views) are being emptied of meaning. And companies that keep measuring their marketing on these indicators are reading thermometers that measure the wrong temperature.

What changes? The value no longer lies in the click but in being cited; in being associated with an industry authority at the moment the AI has to build its answer.

It is a paradoxical return to the fundamentals of pre-internet branding: top of mind, mental availability, brand salience. Concepts that Byron Sharp of the Ehrenberg-Bass Institute codified years ago, and which today suddenly become operational in the most literal sense: if your brand is not “in the mind of the AI”, it is nowhere.

And this is where time, and speed, come back in. AI systems learn. What they learn today about who is authoritative on which subject becomes progressively harder to change. AI citations accrue compound interest, like domain authority in SEO. Those who secured a place first in training datasets, in cited sources and in industry conversations start with a structural advantage that every month of delay amplifies.

To draw an analogy, it is like arriving on the American West Coast in 1849, during the Gold Rush. Those who arrived first took the best mines. Those who arrived in 1851 found land already claimed and had to buy it from the first arrivals. Those who arrived in 1860, by contrast, found an established system in which staying was possible, but getting rich practically was not. GEO in 2026 is California in 1849. GEO in 2028 will be 1860.

AI governance: the problem many Italian companies are about to discover too late

There is a fourth front that, in conversations with Italian CEOs, always comes up last. Not because it is the least important, but because it is probably the least visible. It is the governance of artificial intelligence within the company: who decides which AI tools are used, on what data, for which decisions, with what controls and under whose legal responsibility.

For years this seemed an issue for large multinationals. Today it is becoming an operational issue for any company that uses ChatGPT (even if only) to write emails.

The EU AI Act, in force since 1 August 2024 and being applied progressively, set the deadline of 2 August 2026 for obligations on high-risk AI systems (Annex III: employment, credit scoring, education, access to essential services, law enforcement). Even with the postponement discussed by the European Commission in May 2026 through the Digital Omnibus on AI (which some sources, such as Travers Smith, suggest could push certain obligations to the end of 2027), the strategic direction is crystal clear: anyone using AI in an undocumented, unauditable, ungoverned way will soon find themselves in regulatory proceedings or, worse, in civil litigation with people who feel harmed by automated decisions.

Yet the point here is not the fines. The real issue is the time needed for well-structured AI governance. It requires internal policies, mapping of the systems in use, risk assessment and clearly defined roles (who is the human in the loop). In short, companies that began structuring this governance in April 2026 will reach the summer with something in hand. Those planning to start in July 2026 will reach the end of the year still building, with the regulatory window already open and clients asking to see compliance documentation before signing contracts.

This is the point at which AI governance becomes a commercial lever. Enterprise clients, public administrations and large companies are already adding specific questions on suppliers’ AI governance to their vendor due diligence questionnaires.
Those who answer “we’re discussing it” lose the contract.
Those who answer with a serious document win it.

The fast fish, here, is the one that got itself organised six months before the slow fish. And it shows immediately, from the very first meeting.

The maths of delay: what waiting costs

At this point someone will say what we all say when faced with an irritating pressure: “Fine, I understand, it’s just that I have other priorities right now. I’ll think about it in September”.

Fine. That makes sense. But let us take a moment to look at the numbers behind delay.

Cost one: AI citations are cumulative.

Every month in which your brand is not cited by AI is a month in which the models learn that, on that subject, the authority is someone else. When you decide to move, you will have to spend resources not only on positioning yourself, but on shifting the inertia built up around the competitors who moved first.

Cost two: traditional organic traffic is falling, and it is not coming back.

If your company depends on organic Google traffic for 30%, 40% or 50% of its revenue, every month without diversifying towards AI search, AI-citable content and alternative channels is a month in which you are exposing revenue to a variable you no longer control. In June 2025 the Wall Street Journal documented a 55% decline at Business Insider, with immediate consequences for jobs. These are not isolated cases: this is the pattern across the sector.

Cost three: the cost of customer acquisition rises.

When organic (free) traffic falls, companies try to make up for it with paid. But paid has its own maths: the more companies use it, the higher CPCs climb. Google and Meta know this perfectly well, and it is exactly why they have pushed towards a pay-to-play model. Those who do not build an alternative channel now (visibility in AI) will find themselves, twelve months from now, paying higher CPCs to recover the same revenue.

Cost four: emergency compliance costs three times as much.

Structured AI governance built with twelve months to spare can be done on a normal budget. The same governance, done in three months under regulatory pressure or an urgent request from a major client, costs two or three times as much, because it requires dedicated external consultants, overtime and emergency process restructuring. The same applies to ESG governance, and the same mechanism was already seen with GDPR in 2018: those who moved in 2017 paid a structural cost; those who moved in May 2018 paid three times as much to do the same things, hastily and badly.

Cost five: talent moves elsewhere.

The best people want to work for companies that understand where the world is heading. Companies still talking about a “90-day editorial plan” and “Google Search ad campaigns” as their core strategy have already lost the war for junior talent, and are losing the one for senior talent. Every month of strategic delay is a month in which your competitors are hiring the people who, in three years, will build their advantage.

These five costs are compounding. They multiply one another. A year’s delay does not cost 1: it costs 5 or 6, because each individual cost amplifies the others. This is the arithmetic of the slow fish.

What the fast fish are (actually) doing

But enough about problems, and on to solutions: what exactly are the companies securing first place actually doing?

Five things, in order of urgency.

Thing one: they are auditing their own AI visibility.

The first strategic question today is: “how often is my brand cited by ChatGPT, Perplexity, Claude and Google AI Mode on the queries that matter to my business?”. There are platforms dedicated to measuring this (Profound, Evertune, Omnibound, AI Search and others). The companies moving first carried out this audit in the first quarter of 2026. Those lagging behind are only discovering it now.

Thing two: they are redesigning their content to be cited, not clicked.

The difference lies entirely in the format. Classic SEO content is written to persuade the user to click and read everything. GEO-ready content is written to be extracted in pieces by AI: definitions at the top, structured data, explicit FAQs, authors with verifiable authority. It is an editorial paradigm shift. It is no longer “write at great length for ranking”. It is writing with absolute clarity for extraction.

Point three: they are building presence where AI goes fishing.

Evertune data show that ChatGPT cites Wikipedia, YouTube, Reddit and LinkedIn more than other sources. This does not mean filling Wikipedia with sponsored entries (that would be counterproductive and against its policies). It means that companies wanting to be cited by AI must build an authoritative presence on the channels AI considers authoritative: video content on YouTube, editorial activity on LinkedIn, participation in genuine conversations on industry forums. It is a paradoxical return to content marketing done properly, after ten years of shortcuts.

Point four: they are structuring internal AI governance.

This step means mapping which AI tools are in use across the company, by whom, on what data and for which decisions. Defining who is responsible for oversight. Setting validation criteria for outputs. Documenting everything. In large companies this work takes six to twelve months, in mid-sized ones three to six, and in small ones as little as eight to twelve weeks if carried out with a capable advisor. Companies waiting “for the regulations to become clearer” are using an excuse: the direction of travel is perfectly clear, and those who wait arrive late.

Thing five: they are separating strategy from execution.

This is the point we at Bliss Agency stress most with our clients. Positioning strategy (what you stand for, which topics you want to be an authority on, which conversations you want to be cited in) must be robust, long-term and anchored to the brand. Execution (what you publish tomorrow, where, in what format) must be fluid, responsive to signals and able to exploit new platform features in the first days after release. Companies with this dual speed (steady strategy, fast execution) are the fast fish of 2026. The others are either slow on both (the majority) or chaotic on both (the dangerous minority).

What to do on Monday morning

If you have read this far, you are among those who understand that something needs to be done and are trying to work out where to start. Here are three concrete actions to take in the next seven days, before the next meeting distracts you.

Action one: open ChatGPT (or Perplexity, or Claude, or Google AI Mode) and ask five questions about your sector.

The ones your prospective clients would ask when looking for a supplier like you. Note how many brands are cited in the answer, which ones they are, and whether yours is among them. If it is not, you have just identified the number one priority of your 2026 strategy. Is it there, but only marginally? You have identified priority number two. If it is there prominently, you are already a fast fish, and this article serves to confirm that you are doing the right things.

Action two: call a 60-minute meeting with your marketing director, IT and legal to map how AI is used across the company.

How many AI-based tools are already active in your processes? Who uses them? On what data? Who oversees them? Who is accountable when they get it wrong? You will probably discover two things: that you use far more of them than you thought, and that there is no overall governance framework. That map, even if drawn up in 60 minutes, is the first serious step towards AI Act compliance and towards governance that counts as an asset.

Action three: set a budget and a deadline.

The problem with strategic decisions is that without a budget and a deadline they remain discussions. Decide how much to invest over the next six months in GEO positioning, AI-ready content and AI governance. It does not have to be a huge sum. For a well-structured Italian SME, starting with figures in the order of 30,000-80,000 euros for the first half-year is realistic and produces measurable results. For larger companies, the figures scale proportionally. What matters is the decision to move. Every extra week of indecision is a week that the fast fish, on the other side of the market, is putting to use.

The tide is turning

There is a scene we see repeated in conversations with new clients, and it is worth recounting to close. The entrepreneur or CEO tells us, calmly, that they have always believed in traditional marketing and that, until now, things have worked. But then they sense the waters are shifting: their clients are telling them strange things (“I asked ChatGPT to recommend a company like yours, and it didn’t mention you”), the website no longer attracts the leads it used to, and newer competitors (even smaller ones, even lesser-known ones) suddenly seem to swim faster and hold ground in every single industry conversation.

At that point they ask the question everyone asks: “How much time do I have?”.

The serious answer, based on the data you have just read, is: less than you think.
The new current, which is sweeping over every fish (big and small, fast and slow), is shifting. Right now, as you read, the waters are filling with species that did not even exist before. The fast fish have already changed direction and taken the richest grounds. But the slow fish are still sitting on the shore, wondering whether it is worth following the wave.

The figures from McKinsey, Forrester, Bain, Adobe, Similarweb and Pew Research say so. All the data tell the same story: in technological transitions of this scale, the first-mover advantage is exponential, and the price paid grows proportionally with every month spent delaying a change of pace.

The current is shifting. And like all the great tides that have redrawn the map of the world (Gutenberg’s printing press, electricity, the internet, the mobile revolution), the ecosystem does not wait for the fish that stay still. It keeps moving with those who have started to swim.

At Bliss Agency we work with partners who have understood that it is time to ride this new wave. And our job today is to help entrepreneurs and boards do so. So as not to be left at the bottom of a sea whose rules have changed.

Time is running out. And this time, those who stand still really will be eaten.