For more than twenty years, many digital businesses measured their results based on the clicks they obtained. Appearing on the first page of Google meant visits; visits meant advertising, subscriptions, business opportunities or visibility. The entire architecture of digital marketing \u2014from SEO to media buying, through link building<\/em> and content marketing<\/em>\u2014 was built on a simple premise that was often very hard to refute: the search engine results page was the main artery channelling attention towards the websites of companies and media outlets. If it could be the first page, all the better for being better than the second.<\/p>\n Those of us who work in and around all of this know that this premise is ceasing to hold at an accelerated pace. AI-generated summaries \u2014Google AI Overviews, AI Mode, conversational search engines such as Perplexity or ChatGPT Search, browsers with built-in assistants\u2014 are answering more and more questions without the user needing to leave the interface. The answer is consumed where it is produced, and the visit, that basic unit which ordered two decades of digital strategies, is contracting at a speed the sector’s business models still cannot absorb.<\/p>\n This is not an exaggeration. It is a trend with clear and growing empirical evidence. It is something we see every day on our screens, but it is also a reality that reaches us backed by some of the most rigorous studies published on digital behaviour in recent years.<\/p>\n To understand the magnitude of the change, it is worth knowing how this economy, now subjected to disruption, was built. In the late nineties and early two thousands, search engines transformed a fundamentally chaotic internet into a market ordered by the logic of positioning and keywords. This process was made possible above all by Google, which, from a position of dominance and centrality, ended up being responsible for deciding which pages existed and which remained invisible.<\/p>\n Around that, an entire industry was born \u2014positioning consultancies, analytics platforms, advertising networks, media agencies\u2014 devoted to optimising every step of the journey between search and visit. Basically, SEO (Search Engine Optimization) was born.<\/p>\n The moment the search interface decides to answer any question it is asked by itself, a disruption occurs: content remains necessary, but the visit disappears. This is no minor matter when you consider that in many cases it was the visit that made the creation of the content itself possible. Many things are breaking at once.<\/p>\n A reference point for understanding the phenomenon comes from the Pew Research Center, which in mid-2025 analysed the real browsing behaviour of 900 U.S. adults over a full month<\/a>. We are talking about more than 68,000 Google searches monitored one by one, with access to the terms searched and to the complete browsing history of each participant.<\/p>\n The result was blunt. When an AI summary appeared on the results page, users only clicked on a traditional result in 8% of visits, compared with 15% when there was no summary<\/strong>. In other words, the probability of a click is reduced by almost half. And practically 1% of users clicked on any of the links cited within the summary itself. Pew Research Center’s full analysis is publicly available<\/a>.<\/p>\n There is still more. Twenty-six percent of sessions ended right after viewing an AI summary, compared with 16% on pages without one. Search becomes the end of the journey, not the gateway into browsing. The search engine turns into a space you do not need to leave to get the information you are looking for, with everything that implies.<\/p>\n Around one in five searches already generates a summary, and for queries phrased as questions, 60% of them already trigger one. Precisely those with the greatest informational value are the most affected by this circumstance. The typical summary averages just 67 words, cites three or more sources in 88% of cases, and only about 5% of the links it contains point to news sites. The outlook for anyone depending on search traffic is hardening in a structural way. This is not a fad or a passing circumstance.<\/p>\n SparkToro’s Zero-Click study<\/a> had already documented that in 2024, 58.5% of Google searches in the United States ended without any click to an external website. A randomised field experiment with more than a thousand users, conducted by researchers at the Indian School of Business and Carnegie Mellon University<\/a>, quantified the causal effect: the presence of AI summaries reduces outgoing organic clicks by 39.8% and increases zero-click searches by 34.5%. The Digital Content Next association<\/a>, which brings together premium publishers, documented a median year-on-year decline of 10% in traffic from Google Search among its members during the spring of 2025, with 7% for news brands and 14% for the rest. It is a reality that keeps moving forward, and those of us who keep a close eye on Search Console have been suffering it for a while already.<\/p>\n Behind these figures lies a paradigm shift that deserves to be named precisely. The traditional search engine was above all an index: it ordered alternatives and guided you through the process of navigating between web pages. The new answer engines are conversational interfaces that optimise the research process<\/a>. The synthesis layer enabled by AI is not an add-on to search engines or LLMs; it is their very heart and their reason for being.<\/p>\n Moreover, this answer or synthesis layer does not live only in the search engine: it is also present in browsers with built-in assistants, in messaging apps, in voice assistants and, increasingly, in agents that do not just answer but already act on the user’s behalf.<\/p>\n Every new point of contact between people and the information they consume works according to the same logic: the answer stays in the interface. The click does not disappear overnight, but it is losing territory at forced march.<\/p>\n Google maintains that the clicks that survive are the ones of higher quality. That is in any case its truth, and around this point of view discussions of great relevance are in any case opening up. According to the Reuters Institute’s Journalism, Media and Technology Trends and Predictions 2026 report, publishers fear their search traffic will fall by as much as 43% over the next three years. The report dedicates one of its central chapters to the so-called answer engines and their implications for access to information. The full Reuters Institute report can be consulted on its institutional website<\/a>.<\/p>\n In December 2025, the European Commission opened a formal investigation into whether Google’s AI Overviews and AI Mode use editorial content without sufficient compensation and without adequate opt-out options for publishers. The debate over who pays for the informational value that feeds synthetic answers has ceased to be an industry conversation and has become a matter of public policy.<\/p>\n For news media, the click was the unit that converted attention into advertising revenue and into a commercial argument for core factors of their business such as, for example, selling subscriptions or advertising space. When the click disappears: fewer visits mean less advertising revenue, newsroom cuts and lower demand for quality information. Not only does the media business deteriorate; along the way, the social role of journalism and the quality of the information that reaches citizens deteriorate too. To this we must add the censorious role these platforms acquire. If Google once had the power to hide certain content, now it is the one that interprets reality for you and serves it to you processed.<\/p>\n Newsrooms and agencies that do not have a direct, owned relationship with their audience find themselves, in this scenario, doubly exposed to the design decisions of platforms over which they exercise no control.<\/p>\n For communication agencies and brands, the problem has a different geometry but the same root. If users no longer reach the website, classic search engine optimisation remains necessary but ceases to be sufficient. The new frontier is “citability” in generative engines. It is not enough to rank; you have to be the source that AI systems choose to ground their answers, and you have to mind looking good when you appear.<\/p>\n Consider a query with commercial intent. When a user asks an assistant what the best solution to a specific problem is, the answer names three or four options. If a business is not in that answer, for that user it simply does not exist, regardless of its factual positioning in the market where it operates.<\/p>\n This has concrete implications for all kinds of companies. Content must be structured with data, entities and evidence that models can interpret, you must monitor how AI talks about you, and you must measure presence in synthetic answers with the same rigour with which ranking on Google used to be measured.<\/p>\n The corporate website ceases to be a shop window and becomes a structured knowledge base feeding the layer of synthetic answers offered by search engines or LLMs. Those who fail to care for their website as a digital asset will discover how third parties of all kinds define their positioning.<\/p>\n It may seem like a contradiction, but the reality is that in the face of this trend towards the disappearance of traffic, keeping your own website in perfect shape and working on it actively is more necessary than ever.<\/p>\n There is a third level, less visible but deeper. It is what is being labelled the “feedback loop”. If synthetic answers reduce traffic, and traffic sustains professional journalism, then original, verified content \u2014precisely the material models need to feed on in order to deliver reliable answers\u2014 is weakened. An ecosystem where synthesis thrives while the source suffocates is an ecosystem consuming its own fuel, and no technological improvement will compensate for the progressive loss of the raw material that makes synthesis itself possible.<\/p>\n This risk should not be read only as a business problem. Verified, quality information is a good of collective interest: it sustains the scrutiny of power, social cohesion and the democratic health of communities. That its financing depends on interface design decisions in the hands of a handful of platforms is a legitimate concern.<\/p>\n How is informational value remunerated in the age of synthetic answers? It is quite possible that the answer to this question will not be to our liking.<\/p>\n Faced with this scenario, what can an organisation do to avoid falling behind?<\/p>\n The available evidence would point to five complementary lines of action, valid for media outlets, agencies and businesses alike. None is sufficient on its own; together, they sketch a reasonably robust strategy for an environment that will keep changing.<\/p>\n The first is differentiation<\/strong>. Synthetic answers feed on what is easily “generalisable”. What they cannot synthesise is exclusive access to content, original research, original data and first-hand experience. Generic content turns into a commodity<\/em> and ceases to differentiate anyone. Distinctive content becomes cited raw material. In news media, the context would reinforce the value of field reporting, specialised verticals and material that is hard to replicate.<\/p>\n The second is structure<\/strong>. Being cited by models requires technically readable content: structured data, clear entities, verifiable evidence, clean metadata and well-designed, well-maintained digital assets. “Citability” is largely a problem of editorial engineering, and whoever can keep solving it will hold a competitive advantage over those competitors that do not find the way. It is something hard to imitate in the short term, because a priori<\/em> it depends on accumulated discipline and not on one-off technical fixes.<\/p>\n The third is the direct relationship<\/strong>. If platform intermediation becomes more profitable for the platforms and more expensive for everyone else, a logical response is to try to reduce dependence on it. Newsletters, apps, communities, events… The audience an organisation knows by name and that consumes its product recurrently does not disappear when an algorithm changes or an interface is redesigned. The direct relationship is, today, the only truly owned distribution asset.<\/p>\n The fourth is diversification<\/strong>. Click-based advertising will not return as the support of any serious model. Content licences for model training and retrieval, data products, B2B services, training and events are emerging as complementary revenue pillars for news media and even for news agencies.<\/p>\n The fifth is advanced analytics<\/strong>. If the unit of value migrates from the visit to the answer, indicators must migrate with it: mentions in AI answers, accuracy of representation, brand traffic, assisted conversion or recurrence of the owned audience become new metrics that must be given prominence. Few things are as urgent today as rebuilding dashboards on foundations that reflect current reality<\/a>.<\/p>\n The click will not disappear entirely. Relevant decisions still require visits. Whoever announces the death of the website is just as wrong as whoever denies that something essential has changed. What is at stake is not the disappearance of the click, but its devaluation. It becomes one of several metrics that explain an attention economy that is becoming far more complex.<\/p>\n Nor is it honest to sell artificial intelligence as a cost-free opportunity. It is a formidable technology, which I use intensively and whose professional potential I recognise without reservation, but it is altering the balance of power between platforms and content producers in a way that the market, on its own, is capable of correcting in the interest of societies. User convenience and the sustainability of the information ecosystem do not always push in the same direction, and acknowledging that clearly is the first step to managing the transition without giving up what makes professional information valuable.<\/p>\n The answer economy is, in the end, an economy of attribution. For years, media outlets and brands gave away value in exchange for visibility. Now visibility is often granted without the visit, and value must be claimed through other channels: licences, direct relationships, differentiation and structured presence in the answer layer.<\/p>\n The organisations that understand early that they are no longer competing only for the click, but to be the source that machines talk about when they answer on our behalf, will reach the other side of this transition sooner.<\/p>\n<\/div><\/div><\/div>Twenty years of click economy<\/h2>\n
The numbers behind an announced decline<\/h2>\n
From search engine to answer engine<\/h2>\n
What is at stake: three actors, three urgencies<\/h2>\n
News media and news agencies<\/h3>\n
Communication agencies and brands<\/h3>\n
The systemic risk<\/h3>\n
Competing when the answer no longer drives visits<\/h2>\n
The future that will gradually take shape<\/h2>\n