There’s a version of this conversation happening everywhere right now. Traffic is down. AI answers are replacing search results. Social platforms dominate discovery. And somewhere in a startup Slack, someone is asking whether they even need a website anymore. The claim that websites are dead has started to feel less like provocation and more like received wisdom, which is exactly when it’s worth interrogating.
The short answer is no, websites aren’t dead. But the version most companies built? That one has a problem.
The data that makes the argument feel convincing
The numbers behind the “websites are dead” narrative are real, even if the conclusion is wrong. According to Fortune’s 2025 reporting, nearly 60% of Google searches now end without a single click, and AI Overviews are appearing in roughly 18% of searches, a figure that continues to climb. Clickthrough rates are declining across virtually every industry, and organic traffic to many content-heavy sites has dropped significantly. Digiday has reported on major news publishers watching hundreds of millions of monthly visits evaporate as AI surfaces answers directly.
Alongside this, ChatGPT has reached over 700 million weekly users and processes around 2.5 billion prompts per day. Users increasingly ask rather than browse: they type a question, receive a synthesised answer, and move on without ever landing on a company’s homepage. The interface of the internet has shifted from navigation to conversation, and that shift is structural, not cyclical.
Buyer journeys have also fragmented. Research from Showit’s Are Websites Dead? video suggests buyers typically consult around five sources before making a purchasing decision, with social platforms playing an increasingly dominant role in initial discovery, particularly for Gen Z audiences. Low-ticket purchases are often decided entirely within social ecosystems. For higher-ticket or more complex products, the journey is longer and more research-intensive, but it rarely begins on the brand’s own website.
It’s also genuinely easier than ever to operate without a website at all. Etsy, Gumroad, Substack, and native social commerce tools mean that a creator or early-stage business can generate revenue and build an audience without a single page of their own. This wasn’t true five years ago.
Why this is still a misdiagnosis
The error in the “websites are dead” argument isn’t that it ignores the data, but rather it’s that it misreads what the data means. Lower traffic and fewer clicks don’t mean websites have become irrelevant; they mean websites have changed function.
The most important thing to understand about AI-generated answers is that they are synthesised, not created from nothing. When ChatGPT, Perplexity, or Google’s AI Overviews provide a response, that response is drawn from crawlable web content, from structured pages, published explanations, documented definitions, and indexed sources. SEO analysts including Kaspar Szymanski and Rui Nunes have made this point clearly: remove websites from the equation and AI has no source material to work from. The irony of the “websites are dead” argument is that the thing everyone is looking to instead— AI—runs almost entirely on websites.
Zero-click behaviour doesn’t mean zero need for content. It means the content is being consumed at a different layer of the stack. Users aren’t clicking through, but they are still relying on information, structured explanations, and authoritative sources to form the AI-generated answers they receive. If your company isn’t producing that kind of content, you’re simply invisible, not because no one is searching, but because AI has nothing to say about you.
What actually changed
The more useful framing is not “are websites dead?” but “what is a website for now?”—and the honest answer is that its function has fundamentally shifted.
For most of the web’s history, websites were destinations: the place where discovery, navigation, and conversion all happened in sequence. Users found your site, browsed your pages, and decided whether to engage. That model assumed the website was the beginning of a user’s journey with your brand. In 2026, it’s more likely to be the middle or the end, with discovery having already happened in a feed, a chat interface, or an AI summary.
The homepage has lost its role as the primary entry point, and the traditional navigation model (menus, category pages, linear journeys) matters far less than it did. What matters more is whether your content is structured in a way that allows it to be extracted, interpreted, and surfaced by AI systems before a user ever decides to visit at all.
Christopher Penn and others working in AI content strategy have described this shift as moving from websites as interfaces to websites as infrastructure. The website is no longer primarily a place for humans to browse; it’s a source layer for systems to interpret. It’s a change in role that requires a completely different approach to how content is written, structured, and organised.
The new function of a website in 2026
If websites are no longer destinations in the traditional sense, they are still doing several things that nothing else can replace.
They remain the primary source of truth for what a product is, how it works, and how it compares to alternatives. AI systems use this content to form descriptions, generate recommendations, and answer questions about your brand. A website that fails to define its own product clearly is handing that interpretation over to whatever fragmented, inconsistent information exists elsewhere, which for most companies is a significant risk.
They are also the credibility layer for higher-stakes decisions. Showit’s research reinforces this: for complex, expensive, or trust-dependent purchases, buyers still do deep research, and that research typically leads back to the brand’s own web presence. Social proof, case studies, depth of explanation, and documented methodology all still live most authoritatively on a website, not in a post or a profile.
Websites are also the only owned asset in a landscape of rented platforms. The algorithm risk of building entirely on social infrastructure is not theoretical, it’s a structural dependency that has materially hurt brands when platforms change their reach dynamics. Email lists and owned web presence remain the only parts of a company’s digital footprint that don’t belong to someone else’s platform.
What that restructuring looks like
The shift Rui Nunes and others in the AI visibility space describe is from SEO to a layered model: SEO for discoverability, AEO (answer engine optimisation) for extractability, and GEO (generative engine optimisation) for consistent AI representation. These aren’t separate disciplines so much as a new understanding of how content needs to work across multiple retrieval systems simultaneously.
In practical terms, this means moving away from content designed primarily for keyword ranking and toward content designed to answer clearly and be extracted cleanly. Definition pages that explain what a product actually is. Comparison pages that articulate how it differs from alternatives. FAQ structures that pre-empt the questions AI systems are likely to be asked about you. Internal linking that reinforces consistent entity definitions across the site. Less clever navigation design, more structured explanation.
The companies that will struggle most in this environment are those whose websites were built to impress human visitors with aesthetic complexity but provide AI systems with very little they can actually interpret. The ones that will perform best are those that treat their website as a content architecture problem: a system for making the brand understandable, consistent, and citable.
So what’s the real risk?
The biggest risk for most companies in 2026 isn’t not having a website. It’s having one that AI can’t understand: a site full of vague positioning, inconsistent descriptions, and content that exists but doesn’t explain anything clearly enough to be surfaced or cited.
For Web3 companies, this problem is compounded significantly. Web3 products are already among the most difficult categories for AI systems to interpret accurately, partly because the concepts are genuinely complex, partly because the terminology is inconsistent across the ecosystem, and partly because most Web3 content was written for insiders rather than for structured machine interpretation. When AI can’t understand a product, it either misrepresents it or ignores it entirely, and in a landscape where AI is increasingly the first point of contact between a brand and a potential buyer, that invisibility has real consequences.
Websites didn’t die. They stopped being destinations and became infrastructure: the source layer that AI systems depend on to understand, represent, and recommend the companies operating on the web. The question isn’t whether you need one. It’s whether the one you have is doing its job.
Make your website understandable to AI systems.