Joe Marshall · 9 June 2026
Bot Traffic Has Overtaken Human Traffic Online: What 57.5% Means for Your Website
For the first time, machines browse the web more than we do. Cloudflare says bots now generate 57.5% of requests to web content, against 42.5% from people, driven by agentic AI. Here is what that actually changes for how you build, secure and measure your website.
For the first time in the history of the web, machines are browsing it more than we are. Cloudflare, which sits in front of a large slice of the world's websites, reports that automated systems now generate around 57.5% of all requests to web content, against 42.5% from people. Quietly, bots have become the majority audience online.
Share of HTTP requests to web content, 2026. Source: Cloudflare.
Cloudflare's chief executive, Matthew Prince, admitted the crossover "happened faster than I predicted". He had expected bots to overtake humans in 2027. It arrived a year early.
Year bots overtook human web traffic: forecast versus reality. Source: Cloudflare CEO Matthew Prince, via Tom's Hardware.
The driver is not the spam crawlers and scrapers we have argued about for years. It is agentic AI: autonomous programs that browse the web on behalf of assistants like ChatGPT and Gemini. Ask an assistant to compare ten suppliers, pull pricing or book something, and a single agent might visit thousands of pages to finish a job that would have taken a person a handful of clicks. Cloudflare has gone as far as arguing in a recent post that the old "bots versus humans" framing is now the wrong question entirely.
This is a real shift in who, or what, your website is built for. Here is what it changes in practice, and what we would do about it.
Your analytics are probably already lying to you
If most requests to the web are now non-human, the headline numbers in your reporting deserve a second look. Most analytics platforms filter the obvious, well-behaved crawlers, but the newest agentic traffic is far harder to spot, and some of it executes JavaScript and behaves a lot like a real visitor. The result is that a "great traffic month" can be partly machines doing research, not people who might buy.
Before you celebrate or panic over a spike, separate human engagement from automated activity, and judge performance on the signals that genuinely indicate intent: qualified enquiries, sales, returning users. If your dashboards cannot tell the two apart, that is the first thing worth fixing. This is exactly the kind of clean, trustworthy measurement we build through our analytics work.
Build content an AI agent can actually read
When an agent visits your site, it does not see your brand. It does not register the hero video, the carefully art-directed photography or the clever hover animation. It reads structure. Clean semantic HTML, content that is present without waiting on heavy client-side rendering, sensible headings, clear product and pricing information, and structured data all make the difference between being understood and being skipped.
If you want your business to be the one an assistant cites or recommends, the page has to be legible to the machine doing the reading. That is the heart of GEO and LLM optimisation, and it sits right alongside the fundamentals of good SEO rather than replacing them. The brands that win the next few years will be the ones that are equally readable to a person and to an agent acting on their behalf.
Decide which crawlers you actually want
Not every bot is equal. Some agents are the very thing that lands you a citation inside ChatGPT or a recommendation in Gemini. Others are scraping your content to train a model, or hammering your server for no return. Blocking everything is easy and usually a mistake: shut out the wrong crawler and you can quietly remove yourself from the AI answers your customers are starting to trust.
Tools like robots.txt, edge controls at your CDN, and Google's new opt-out toggle for AI Overviews give you real levers here. The point is to use them deliberately. Decide which automated visitors earn their access and which do not, rather than leaving it to a default nobody chose. It is a strategic decision about visibility, and it should be treated as one.
Do not forget the humans who still buy
For all the talk of agents, people still hold the credit card. Increasingly the journey looks like this: an assistant does the research, narrows the field, and then a human arrives on your site to make the final call. So you are now designing for two audiences at once. The machine has to be able to parse and shortlist you, and the person it sends has to land on something fast, clear and convincing enough to act on.
Over-rotate towards the bots and you end up with a site that ranks in AI answers but does not convert the humans who follow. The goal is not to choose between them. It is to be excellent for both.
The takeaway
The majority of your website's visitors are no longer human, and that is not a temporary blip. It is the new shape of the web. The businesses that adapt will treat their site as something an AI agent reads and a person buys from, measure the two separately, and make a conscious choice about which machines they let in.
If you want a clear view of how AI agents are interacting with your site, what your analytics are really telling you, and how to stay visible in AI answers, talk to us. It is the sort of structural shift we think about every day.