AI is reshaping web discovery, forcing brands to optimise for machines and trusted third-party sources to stay relevant.
THERE is a growing theory circulating across marketing, media and technology circles that the public web is heading towards extinction.
The logic is simple enough. If AI can answer questions instantly, compare products on your behalf, summarise reviews and even complete transactions for you, why would anyone still visit a website?
According to institutional consensus from Gartner, McKinsey, Bain and Reuters Institute, organic traffic is expected to contract by 15% to 50% by 2028. While the direction may be right, the conclusion is probably not.
The public web is unlikely to disappear. What is disappearing is the version of the web that marketers have optimised for over the past two decades.
Compression of the funnel
The strongest argument for the collapse of the public web is that AI fundamentally changes the mechanics of discovery. Historically, search worked as a link economy.
For one small business owner I know, the impact reality hit home when the leads her website had generated for more than a decade suddenly dried up. The discovery layer had shifted towards AI but her website was not structurally designed for AI to access and extract its content.
AI is completely changing digital lead generation. Instead of directing users to a list of websites or urls, it delivers compiled information from multiple sources.
The result is a dramatic decline in referral traffic. Today, around 60% of Google searches end without a click and it is worse when it is AI-led. Google’s AI-mode telemetry reports zero-click behaviour over 90% of the time.
The most extreme vision of a future in which search is bypassed altogether is what the industry calls the “agentic takeover”. In this scenario, AI agents interact directly with websites on behalf of users, collapsing the customer journey from discovery to procurement in a matter of moments.
To remain relevant, websites will need to deliver highly structured, hyper-personalised content that AI agents can interpret, extract and act upon for specific users.
In many categories, especially sectors like finance, AI systems disproportionately cite publishers, forums, aggregators and third-party sources over the brands themselves. This is because the brand still optimises their website for search instead of AI-generated answers.
Data shows that visitors referred by AI can convert at significantly higher rates than those from traditional organic search – by as much as 23 times in some cases. This is because AI can pinpoint exactly what the user is looking for and gives them exact references to suit their needs. As a result, overall website traffic may shrink but that does not diminish the website’s value. Instead, it receives far more highly targeted visitors.
Designing for two audiences
Today’s websites must be designed for two distinct audiences: the human-seeking validation and the machine-seeking structured information. In June 2026, Cloudflare reported that, for the first time, machines were discovering and consuming HTML content at a higher rate than humans.
To meet the requirements for AI-driven discovery demands a fundamental overhaul of your content and the website’s formatting.
You need to clearly articulate your proposition because machines do not interpret ambiguity well. That implication is profound: the brands that succeed in AI discovery will not necessarily be the loudest but the clearest, with information that is structured for machine legibility.
The strategic question is no longer: “How do we get more clicks?” It is: “How do we become the source the AI trusts by default?” Each approach is different.
Will public web traffic go to zero?
The probability of public web traffic devolving to zero remains extremely low. It won’t vanish entirely because humans still need trust, validation, accountability and transactions anchored somewhere tangible.
What will change is the role of the website and the marketing methods traditionally built around optimising the website for search. Instead of creating websites primarily for people to find and browse, content is increasingly being structured as text and tabulated data, stored in back-end databases that are invisible to human visitors but readily accessible to AI systems.
Path forward
Strategically, companies now need to be super consistent with the messaging about their brands, across platforms and various interfaces.
AI models do not take your marketing copy at face value. They weigh what credible third parties say about you across independent forums, news outlets and peer reviews. Your media is viewed as raw data and your earned authority is what the model trusts.
Information on your business – on websites and other platforms – must now be extractable as machine-readable chunks. AI seeks out so-called “canonical” information (the ultimate source of verifiable truth) about you to report as a baseline.
Regulatory correction
A key obstacle to this evolving approach to communication is regulation. If governments conclude that the energy required for AI to answer even a simple query – such as “What is a current account?”– is unjustifiably high, they may introduce guardrails governing how AI can be used.
Another potential barrier is the emergence of a coordinated global copyright regime that prevents technology companies from treating the open web as a free training corpus. Such a shift would rebalance power away from information aggregators and encourage users to return to the original sources of information.
Until such guardrails materialise, businesses need to feed the machine to maintain baseline visibility. They must also invest heavily in the aspects of their business that AI can never replicate or access.
Shaad Hamid is the general manager of GrowthOps Singapore. Comments: [email protected]
