Your Digital Strategy Was Designed for an Internet That No Longer Exists

For two decades, human attention was the currency of the web. Automated bots now generate the majority of web page requests. The strategies built around earning human visits are operating on assumptions that the data no longer supports.

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://open.substack.com/pub/johanosteyn/p/your-digital-strategy-was-designed?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

On 3 June 2026, Matthew Prince, CEO of Cloudflare — a company that sits in front of roughly one in five websites on Earth and processes a significant portion of all global web traffic — posted a single statistic on X that probably deserved more attention than it received. Automated software bots now generate approximately 57% of the HTTP requests made to web pages worldwide. Human beings account for the remaining 43%. For the first time in the history of the internet, the majority of the activity happening on the machine-readable web is generated not by people browsing, reading, and buying, but by software crawling, extracting, and automating.

Prince noted that he had not expected this moment until late 2027. He had moved his estimate to early 2027. Then it simply happened — months ahead of even his revised schedule, driven by a new kind of bot multiplying far faster than anyone had modelled. “Welp, that happened faster than I predicted,” he wrote.

The digital strategies that most organisations are running were designed for a different internet. The data says that the internet no longer describes the majority of the web.

CONTEXT AND BACKGROUND

To understand what the bot majority means, it is necessary to be precise about what it measures — and what it does not. The 57% figure applies to HTTP requests to web pages: the transactional signals that pass between browsers and servers every time a page is loaded, a link is followed, or a resource is retrieved. It does not measure total internet bandwidth, where a single human streaming a four-hour film consumes vastly more data than a bot executing thousands of requests. By time spent and by data volume, humans remain a significant presence on the internet. What has tipped past the halfway point is the plumbing of the web — the constant background traffic of requests and responses that keeps the system running. In that layer, machines now outnumber people.

The commercial significance of that distinction requires understanding the arrangement the web was built on. For approximately two decades, the internet ran on what the transcript rightly characterises as a handshake. Websites created content worth reading. Search engines crawled that content and pointed people toward it. People arrived, read, and in doing so generated the advertising revenue, brand authority, and commercial conversions that made the creation of content economically viable. Traffic was the currency, and the exchange worked because the human visit was the unit of value that the entire system was calibrated around.

That exchange has been broken at its most critical link. When a person searches Google and an AI-generated overview appears at the top of the results page, they receive an answer synthesised from the content of multiple websites — and in a significant proportion of cases, they close the tab without visiting any of them. The Pew Research Center studied this in March 2025, tracking the actual browsing behaviour of 900 US adults across tens of thousands of searches. When an AI summary appeared, users clicked through to a traditional search result in 8% of visits. When no summary appeared, they clicked nearly twice as often, at 15%. The source links embedded within the AI summary itself — the citations that nominally credit the content the AI drew on — were followed in approximately 1% of all visits.

For every publisher, media company, and content-producing organisation that built its digital presence around search visibility, that is not a trend to monitor. It is an existing hole in the accounts. One firm tracking more than 2,500 news sites found Google referrals down roughly a third over the past year. Business Insider watched its search traffic fall by more than half and laid off a fifth of its staff. Chegg, now suing Google, documented a 49% in non-subscriber traffic as AI began answering the questions that used to bring students to its pages. Penske Media, the company behind Rolling Stone and Billboard, became the first major American publisher to sue Google over the practice, arguing in its complaint that Google has shifted from a search engine that sends users to websites into an answer engine that gives them no reason to leave.

INSIGHT AND ANALYSIS

The bot majority and the AI overview problem are not separate phenomena. They are two expressions of the same structural shift in how the web works. In both cases, the human visit that used to be the unit of value is being intercepted by software before it completes. The bot generates the request but produces no revenue. The AI overview delivers the answer but sends no reader to the source. The surface of the web looks exactly as it always has. The commercial architecture underneath it has been reorganised around a different kind of actor.

The most direct implication for digital strategy is the one that most organisations have been slowest to process. Search engine optimisation — the discipline of making content discoverable and appealing to human searchers — was built for a web where the human searcher was the primary audience for web content. That assumption is now incorrect in a precise and measurable sense. The primary audience for web content is increasingly the AI system that reads, processes, and summarises it before any human sees it. The question of whether your content ranks in a traditional search result is being superseded by the question of whether your content is cited in an AI-generated answer — and whether that citation translates into any commercial value for the organisation that produced the content being summarised.

The legal dimension of this shift is beginning to resolve in ways that have governance implications for every organisation that publishes content and every organisation that is described by AI-generated summaries. A regional court in Munich ruled in May 2026 that when an AI system rewrites source material in its own words and reaches its own conclusions — in the specific case, attributing scams and dubious business practices to named organisations that the cited sources never described in those terms — the search provider is no longer a librarian pointing to existing pages. It has become the author of a new statement, and authors bear liability for false statements. Google is appealing the ruling. The reasoning, however, travels directly into every jurisdiction where an AI-generated summary has made a false claim about a named entity — including South Africa, where defamation law imposes comparable liability on the publishers of untrue statements that damage reputation.

The infrastructure layer of the web is responding to the commercial breakdown with a mechanism that would have seemed far-fetched even two years ago. Cloudflare has deployed pay-to-crawl functionality using HTTP error code 402 — a code that has existed in the web’s design since the early 1990s, labelled payment required, and gone essentially unused until now. When an AI crawler arrives at a site using this system, the site can respond with a price. Pay and the content is accessible. Decline and it is not. Cloudflare now blocks AI crawlers by default on new sites registered through its platform. The architecture of the web, at the infrastructure level, is beginning to treat AI systems as the primary audience — not visitors to court but software to charge at the door.

IMPLICATIONS

For boards and executives responsible for digital strategy, the bot majority milestone and its surrounding developments carry three specific implications that most current strategy frameworks have not yet absorbed.

The first is about audience definition. The marketing, content, and digital investment decisions most organisations are making are calibrated around reaching human audiences through search and social discovery. The data says that the majority of the requests those investments generate are now made by machines, not people. The measurement frameworks — page views, sessions, click-through rates, conversion funnels — were designed to count human interactions. They are not designed to distinguish a human visitor from an AI agent that is infinitely faster, more precise, and leaves no advertising signal or conversion opportunity behind it. The organisations that understand their actual human audience is now a minority of their measured traffic will make better investment decisions than those that treat all traffic as equivalent.

The second is about content strategy. The twenty-year assumption that useful content earns human visits through search visibility is being replaced by a different dynamic: useful content earns citation in AI-generated answers, and the commercial value of that citation depends on whether being named in an AI summary translates into the kind of human engagement that produces revenue. For some organisations and some queries, the answer is yes. For most, the evidence from Pew’s research — 1% of visits result in a click on a cited source — suggests the economics of the old arrangement do not automatically transfer to the new one. The organisations redesigning their content strategy around being cited in AI answers rather than ranked in search results are the ones building for the internet that currently exists.

The third is about reputational risk. The Munich court ruling establishes a principle with broad applicability: when an AI system makes a false statement about a named organisation in its own synthesised language, the operator of that system may be liable for the harm the false statement causes. For every organisation that has been incorrectly described, mischaracterised, or falsely associated with negative attributes by an AI-generated answer — and the number of such organisations is not small — the question of what recourse exists and how to pursue it is no longer purely hypothetical. It is a current and active governance question that most organisations’ legal and communications functions have not yet addressed.

CLOSING TAKEAWAY

The internet is not dying. That framing misses what is actually happening. The web is busier than it has ever been — more requests, more data, more activity across every measurable dimension. What has changed is who or what is generating most of that activity. Bots make the requests. AI systems write the answers. Agents complete the transactions. The humans who used to occupy the centre of that system are drifting toward its edges, receiving finished products and AI-generated summaries of content they no longer visit in the volume that made producing that content economically viable.

The digital strategies most organisations are running were designed for the web that existed before June 2026. The web that Cloudflare’s data describes is one where the assumptions those strategies rest on — human traffic as the primary output of useful content, search visibility as the primary driver of digital presence, the visit as the unit of commercial value — are no longer accurate descriptions of how most of the web works. The organisations that update their strategies for the internet that currently exists will be better positioned than those that continue optimising for the internet that no longer exists.

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net