In July 2026, LBG Media, the company behind LadBible, described the kind of platform shock digital publishers have learned to fear. Its shares had plunged after a second profit warning in two months. The company said indirect revenue, including social media revenue and advertising from its websites, had fallen 41 percent after Meta changed the way Facebook and Instagram promoted publisher content.
This was not a Google AI story. That is what made it useful.
Meta’s change was distribution throttling: the platform still had readers, but sent fewer of them to publishers. AI search is disintermediation: the platform can satisfy the reader before a visit is needed at all. The mechanisms are different, but the dependency is the same. A publisher can build the audience, hire the staff, produce the material, and still depend on a route to the reader that someone else controls.
AI search pushes that dependency into a harsher phase. The platform no longer has to send the reader away.
For most of the modern web, a website was a destination. A publisher wrote the article. A shop built the catalogue. A travel site collected the guide. A recipe blog tested the dish. A forum preserved the lived experience. Google, Facebook, or another platform helped people find it. The user clicked. The site received the visit. Attention became advertising, subscriptions, reputation, leads, bookings, or sales.
That bargain made the open web economically possible. It was never equal, and it was never stable. Platforms always held more power than the sites that depended on them. Still, the exchange had a working logic: websites supplied information, platforms supplied discovery, and users moved between them.
The zero-click economy breaks that movement.
A zero-click internet is one where the answer arrives before the visit. Search engines summarize the page. AI assistants synthesize the explanation. Social platforms show the useful fragment inside the feed. The browser, chatbot, or search result becomes the place where the user’s need is satisfied.
The website remains in the system, but its role changes. It becomes source material.
That is the real break from ordinary search. Traditional search organized attention around links. AI search consumes, compresses, and presents information as a finished answer. The site may still be cited. It may still be credited. But credit is not traffic, and traffic is what pays the bills.
The evidence is no longer theoretical. A 2026 study by Haofei Xu, Umar Iqbal, and Jacob M. Montgomery examined more than 55,000 trending Google queries across 19 categories. It found that AI Overviews appeared in 13.7 percent of queries overall, but in 64.7 percent of question-style searches. That distinction matters because question searches are the heart of much of the informational web: how to fix something, what to buy, where to go, what happened, what a policy means, whether a symptom is serious, which product is better.
Those are exactly the queries where a summary can replace the visit.
A separate 2026 study by Mehrzad Khosravi and Hema Yoganarasimhan looked at Wikipedia and found that exposure to Google AI Overviews reduced daily traffic to affected English Wikipedia articles by about 15 percent. Wikipedia is not a fragile content farm. It is one of the most useful public knowledge projects online. If AI summaries can reduce visits there, smaller publishers, expert blogs, local guides, and independent reference sites have less room to pretend this is someone else’s problem.
The honest limit matters. This is not yet the death of search traffic. In the first study, most trending queries still did not produce an AI Overview. The same research found that most AI Overview claims were supported by the cited pages. AI search is not just a machine for stealing wrong answers from websites. It is often useful, convenient, and good enough for the user’s immediate purpose.
That is why the shift is so powerful.
Bad products announce themselves. Useful products quietly change the system around them.
For users, the zero-click economy feels like progress. Nobody misses opening five pages full of pop-ups, autoplay video, affiliate clutter, cookie banners, and padded paragraphs before finding the answer. The web trained users to distrust the journey. AI search promises to remove the journey.
A parent checking a school deadline, a traveler looking up a visa rule, a shopper comparing two devices, or a patient trying to understand a medical term is not thinking about the economics of the page. The user wants the answer. The site’s business model is invisible at the moment of need.
The problem is that pages are where the work was financed.
Advertising needs impressions. Subscriptions need habit. E-commerce needs the product page. Local services need the form, the call, the booking, the menu, the map, the review trail. Publishers need readers to arrive often enough that a relationship can form. When the answer layer satisfies the user first, the source may still be useful, but it becomes economically distant from the value it created.
This is the cleanest way to understand the zero-click economy: usefulness is being separated from visitation.
That separation rewrites search strategy. For years, websites were told to optimize for ranking. Be clear. Be authoritative. Answer the query. Structure the page. Earn links. Improve speed. Build topical authority. The reward was visibility, and visibility usually meant clicks.
Now a site can be visible inside an AI answer and still lose the visit. It can be cited without being read. It can provide the fact that completes the summary while the platform captures the user’s attention, the advertising space, and the next question.
Being found is no longer the same as being visited.
The pattern is easiest to see in publishing, but it does not stop there. A restaurant’s hours can be answered without opening its site. A hotel’s policy can be summarized without visiting the booking page. A software company’s documentation can solve a problem inside an assistant. A school, clinic, law firm, or public agency can have its explanation absorbed into a response that keeps the user elsewhere. These examples are not separate traffic studies. They are the everyday extension of the same mechanism: the more clearly a website answers a question, the easier it becomes for an intermediary to answer that question on its behalf.
There is also a deeper editorial shift. Search results once forced users to move through sources. That had problems, but it preserved some visible relationship between answer and origin. AI search reduces that friction by turning many sources into one response. The platform decides which material matters, how it is combined, what is left out, and which citations appear.
Xu, Iqbal, and Montgomery found that nearly 30 percent of AI Overview-cited domains did not appear in the ordinary first-page results. They also found that 11 percent of atomic claims in AI Overviews were unsupported by the cited pages. Those numbers do not prove the system is broadly unreliable. They show that the answer layer has its own logic. It is not simply a shorter path to the same web. It is a new editorial surface with enormous power over what users see and what sources receive attention.
A third 2026 study, by Riley Grossman, Songjiang Liu, Michael K. Chen, Mike Smith, Cristian Borcea, and Yi Chen, reached a related conclusion from another angle. Comparing Google Search, Google AI Overviews, and Gemini, the authors found that generative search retrieves and presents sources differently from traditional search. They also found that websites blocking Google’s AI crawler were significantly less likely to be retrieved by AI Overviews, even when the content itself remained available.
That is the trap in miniature. Let the crawler in, and the site may be summarized without a visit. Keep the crawler out, and the site may become less visible in the new layer of discovery.
That power is already producing a fight over access.
Cloudflare’s chief executive, Matthew Prince, has warned that publishers face an existential threat if AI systems consume their work without sending enough traffic back. Cloudflare has pushed tools that let websites block or control AI crawlers. Really Simple Licensing, first announced in 2025 and later released as a formal specification, offers another response. Backed by companies including Reddit, Yahoo, Medium, Quora, O’Reilly, and wikiHow, it extends the old robots.txt idea from access instructions into licensing terms. A publisher can signal not only whether a crawler may enter, but what kind of compensation or permission is expected.
That change matters because it shows the web trying to price what it once gave away through openness. The early internet grew because pages were available to crawl, link, quote, index, and discover. Openness created reach. Reach created revenue. Now openness can feel like exposure. If a crawler can take the useful part and leave the visit behind, every website has to ask a question that once sounded hostile to the spirit of the web: who is allowed to read us at machine scale, and what do they owe?
There will not be one answer. Large publishers will pursue licensing deals. Some sites will block aggressive crawlers. Some will accept the loss as the price of remaining discoverable. Some will chase “generative engine optimization,” hoping to be named inside AI answers even if the click rate falls. The strongest players have leverage because they own brands, communities, databases, tools, marketplaces, or direct audiences. The weakest position belongs to the ordinary useful website: good enough to be mined, not powerful enough to be paid.
Websites will not disappear. They will remain necessary for transactions, archives, identity, community, accountability, and depth. But many informational pages will lose their old economic function. They will still explain the world. They may not receive the reader who asked for the explanation.
The winners will be the sites that are not merely visited for answers, but chosen for trust. The difference is small in language and enormous in business. A commodity answer can be summarized. A trusted relationship has to be built.
This is why small businesses and creators face a colder internet. A website alone may no longer be enough. The page has to become part of a larger relationship: email list, brand habit, booking system, community, product, service, tool, or voice. The open web used to reward being available. The zero-click economy rewards being specifically wanted.
Search tactics alone cannot solve this. The issue is leverage. When discovery, summary, and user attention all sit inside the same platform, the independent website loses bargaining power. It can be included, ignored, cited, scraped, demoted, summarized, or replaced by interface changes it did not choose.
The website was once the front door of digital life. Now it is becoming the kitchen. The work still happens there, but the meal is served somewhere else.
That image is uncomfortable because it preserves the truth on both sides. The internet still needs websites. AI systems need fresh reporting, expert pages, product data, reviews, forums, documentation, images, local details, and human judgment. A web without original sites would eventually become stale, circular, and less trustworthy.
But needing the work is not the same as rewarding the worker.
That is the structural signal. The open web was built on movement: search, click, read, return. The zero-click economy is built on capture: crawl, synthesize, answer, retain. It is easier for users, stronger for platforms, and dangerous for anyone whose business depends on the visit.
The next internet may contain more pages than ever. It may look abundant from the outside. But behind that abundance, a quieter question will decide who survives.
Can a website still matter when the answer no longer has to send anyone there?
Sources consulted: The Guardian on LBG Media and platform traffic shocks; Axios on Cloudflare and publisher risk; The Verge on Really Simple Licensing; The Verge on RSL 1.0; Haofei Xu, Umar Iqbal, and Jacob M. Montgomery, “Measuring Google AI Overviews: Activation, Source Quality, Claim Fidelity, and Publisher Impact,” arXiv:2605.14021; Mehrzad Khosravi and Hema Yoganarasimhan, “Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia,” arXiv:2602.18455; Riley Grossman, Songjiang Liu, Michael K. Chen, Mike Smith, Cristian Borcea, and Yi Chen, “How Generative AI Disrupts Search: An Empirical Study of Google Search, Gemini, and AI Overviews,” arXiv:2604.27790.
