Bot visitors to websites have today crossed a threshold that analysts had long anticipated but few expected to arrive so quickly: automated traffic now accounts for the majority of all web requests, according to the latest data from Cloudflare, while a separate debate sharpens over whether appearing in AI-generated answers translates into any measurable commercial value for the businesses being mentioned.

The numbers that crossed a threshold

Cyrus S., who previously led global SEO at a major tech company and now publishes commentary on search and AI growth, shared today the figures from Cloudflare Radar that have drawn significant attention across the marketing community. The data, covering a seven-day window ending June 5, 2026, shows bots accounting for 57.4% of web traffic to HTML content, with human visitors at 42.6%. The crossing of the 50% threshold, according to Cyrus S., happened "much faster than expected."

The figure itself is striking, but a breakdown shared in the discussion thread by Loganix, a digital marketing company tracking AI crawl patterns, reveals the more consequential split within that bot majority. Of the automated traffic recorded, training-related bots account for 50.6% of the total share, while search-related bots make up just 10.7%. That asymmetry - crawling to train AI models versus crawling to index content for search - sits at the center of a wider tension between content publishers and AI platform operators.

Cloudflare's network sits at a position of unusual visibility into these patterns. The company proxies and protects traffic for millions of websites globally, making its Radar dataset one of the most representative available for understanding internet traffic composition at scale. PPC Land documented in December 2025 that AI bots had already reached 4.2% of all HTML requests across Cloudflare's network in 2025, with human visitors at 43.5% and non-AI automated systems accounting for 47.9%. The jump from that baseline to the figures now circulating suggests the composition has shifted materially within roughly six months.

Training versus search: a lopsided ratio

The distinction between training crawlers and search crawlers matters enormously for publishers trying to understand who is actually consuming their content and why. A search bot visiting a page creates at least a theoretical pathway to a user clicking through from a search result page. A training bot visits for the purpose of extracting content to improve an AI model - a transaction that generates no referral traffic, no attribution, and no revenue for the publisher.

PPC Land covered data in September 2025 showing training-related crawling had reached nearly 80% of all AI bot activity at that point, up from 72% one year earlier, while search-related crawling had fallen to 18% and user-initiated actions made up just 2%. The new Cloudflare figures, at 50.6% for training across all traffic, represent a different measurement methodology - comparing against total traffic rather than only within AI bot activity - but point toward the same structural imbalance.

GPTBot, operated by OpenAI, grew its share of AI crawling traffic from 4.7% in July 2024 to 11.7% in July 2025, according to Cloudflare data reported by PPC Land in April 2026. An analysis of more than 7 billion server log entries published in April 2026 found OpenAI's automated crawlers tripled in activity following the launch of GPT-5 in August 2025, even as data from the same period suggested the number of people directly using ChatGPT may be declining.

Bytespider, operated by ByteDance, has also emerged as one of the most active bots in terms of request volume. The sheer scale of these operations - dispatched by companies whose revenue does not depend on sending traffic back to the sites they crawl - creates what Cyrus S. described bluntly as "a tax on your website." As he put it: "While Google/AI companies monetize all your data, you still pay the hosting bill for them to crawl it."

Rand Fishkin, co-founder of SparkToro and co-author of the forthcoming book Zero Click Marketing, captured the underlying economics in a comment beneath the data post: "Site owners: 'So... you're gonna steal my stuff, republish it on your own site without trademark/copyright exposure, and I'm going to PAY for you to do it?' Google & AI Tools: 'Yup!' Site owners: 'Sigh... OK, fine.'"

Goran Majic, a journalist and SEO analyst, offered a different explanatory angle, asking whether the growth itself was partly self-reinforcing: "More human chatbot usage → more requests → more bot visitors?"

The cache problem and infrastructure cost

The training-heavy crawl pattern carries consequences beyond the conceptual. PPC Land documented in April 2026 that Cloudflare and ETH Zurich researchers identified three characteristics of AI crawler traffic that actively undermine the caching systems websites rely on to manage load. AI crawlers issue high-volume, parallel requests. Rather than concentrating on popular pages - the behavior that cache systems are optimized to handle - they access rarely visited content in sequential, complete site scans. And because they launch independent instances without shared session state, each instance may register as a new visitor, even when requesting identical content.

That behavior defeats what Cloudflare calls its "least recently used" caching policy, which evicts content that has gone longest without being requested whenever storage space runs low. When training crawlers systematically scan low-traffic pages, they push popular content out of cache, increasing origin server load and cost for website operators. The result is that publishers pay for hosting, pay for the increased infrastructure cost driven by training crawls, and receive nothing in return.

Cloudflare's commercial response to this situation has been the development of multiple tools. In June 2024, the company introduced blocking tools allowing operators to prevent AI scrapers from accessing their content. In July 2025, it launched a pay-per-crawl service in private beta, detailed by PPC Land at the time, which creates a mechanism for content creators to charge AI companies for access rather than simply blocking them. Cloudflare CEO Matthew Prince outlined three scenarios for the web's AI future in September 2025, including content creator extinction as one of the possible outcomes.

Brand mentions without conversions

A second discussion gaining traction in the same professional networks focuses on a specific claim that vendors and platform operators have pushed throughout the past two years: that appearing in AI-generated answers - even without a click - carries commercial value because it builds brand awareness.

Himanshu Sharma, who writes and trains on GA4, BigQuery, and voice AI, challenged that position directly in a post on June 3, 2026. The argument rests on observable outcomes rather than theoretical frameworks. According to Sharma: "brand mentions in AI results have no commercial value, despite what vendor-sponsored studies (touting 'visibility = value.') say."

The logic is empirical. If AI visibility were compensating for the organic traffic that search features like Google AI Overviews have displaced, the commercial outcomes - leads, sales, form fills - should hold steady even as traffic falls. That is not what many operators are reporting. As Sharma put it in the comments of the post: "If AI mentions truly compensated for lost clicks, we'd see stable or rising leads/sales despite traffic drops. In reality, many (probably most) sites experiencing significant declines in organic traffic also see corresponding drops in leads and sales. The indirect benefits from AI visibility don't offset the volume loss for a large portion of businesses."

Sharma's position cuts against a large category of vendor research that has emerged over the past 18 months, including studies suggesting AI search visitors convert at higher rates and that brand mentions in AI responses prime users for future purchases through branded search or direct navigation. A Datos-SparkToro report covered by PPC Land in May 2026 showed zero-click searches at their lowest recorded level in Q1 2026, yet with AI tools still below 2% of total desktop visits, meaning the lost clicks are not migrating to AI-driven engagement in any measurable volume.

Ayoub OUARAIN, a senior SEO and GEO consultant, raised a counterexample in the same thread - a reported case from a ClickUp CMO suggesting the company had "nuked" its organic traffic with scaled AI content while leads and paid users from organic were actually higher. Sharma's response was concise: "deinvestment in geo."

Context: what the traffic data shows

The broader trajectory makes the debate over brand mentions more consequential. Small publishers lost 60% of search traffic in the period covered by Chartbeat data through early 2026, with Google Search falling to 5.8% of global page views by February 2026 - roughly half its share from the start of 2024. ChatGPT accounts for just 0.02% of total publisher traffic by the same data source.

AI bots were found to crawl retail sites 198 times more per visit than Google in a March 2026 report by Retail Economics, AWS, Botify, and DataDome. AI-driven bot traffic reached 5.4 times the Q1 2025 baseline across retail websites analysed by Botify by Q4 2025. Meanwhile, 80% of websites were found exposed to agent spoofing, where bots misrepresent their identity to bypass access controls.

HubSpot's April 2026 launch of an Answer Engine Optimization tool came with a disclosure that organic traffic for HubSpot customers had fallen 27% year-over-year. Zero-click searches on Google climbed from 56% to nearly 69% since AI Overviews launched in May 2024. The Google Network revenue line - covering AdSense, AdMob, and Ad Manager - fell 4% year-over-year to $6.97 billion in Q1 2026.

A study covered by PPC Land documented that in German search, the click-through rate at position one falls from 27% to just 11% when an AI Overview appears, amounting to 265 million lost organic clicks per month across 100 million keywords. Ahrefs research found that by February 2026, AI Overviews correlated with a 58% reduction in click-through rates for top-ranking pages - nearly double the 34.5% figure the same company measured in April 2025.

These figures form the backdrop against which Sharma's argument about brand mentions should be read. If a site loses 50% of organic traffic and sees leads fall in parallel, being mentioned in an AI answer that did not generate a click produces no measurable business output. Sharma's position is not that AI visibility never matters - it is that the current evidence does not support the claim that mentions compensate for lost volume, and that conclusions should be drawn from first-party data rather than vendor-commissioned studies.

What this means for digital marketers

For marketing teams managing paid and organic budgets, the data points create a set of practical problems. Bot traffic inflates raw request counts without producing ad impressions, conversion events, or any other signal a campaign manager can optimize against. Microsoft Clarity introduced Bot Activity tracking in January 2026 specifically to help operators understand which automated systems access their properties and how that activity affects analytics data. Without such controls, engagement metrics, session data, and conversion rates become harder to interpret.

The split between training-heavy and search-lite crawling is particularly significant for businesses that have invested in content strategies premised on the idea that strong rankings translate into AI citations, which translate into traffic. The Cloudflare Radar data makes clear that the largest component of automated traffic arriving at most websites has no search intent whatsoever.

The debate over GEO - generative engine optimization - reflects this tension directly. Vendors have built entire product categories around measuring and improving AI brand mentions. Sharma's argument, and the data pattern supporting it, suggests those investments should be stress-tested against actual conversion and revenue data rather than visibility metrics. As the Cloudflare numbers today confirm, a bot visiting a site to train a model leaves no commercial trace at all.

Timeline

  • June 2024 - Cloudflare introduces tools allowing website operators to block AI scrapers and crawlers
  • July 2024 - GPTBot accounts for 4.7% of AI crawling traffic, according to Cloudflare data
  • October 2024 - Google AI Overviews roll out to more than 100 countries
  • April 2025 - Ahrefs finds AI Overviews correlate with 34.5% reduction in organic click-through rates; PPC Land coverage
  • July 2025 - Cloudflare launches pay-per-crawl in private beta; PPC Land coverage
  • July 2025 - GPTBot's share of AI crawling traffic reaches 11.7%, up from 4.7% in July 2024
  • July 2025 - Zero-click searches on Google reach nearly 69%; PPC Land coverage
  • August 2025 - GPT-5 launches; OpenAI's automated crawlers triple in activity
  • September 2025 - Training-related crawling reaches nearly 80% of all AI bot activity; PPC Land coverage
  • November 2025 - Ahrefs finds organic CTR for AI Overviews queries drops from 1.76% to 0.61%; PPC Land coverage
  • November 2025 - UK website traffic growth collapses 86% since Google AI search rollout; PPC Land coverage
  • December 2025 - Cloudflare annual review shows AI bots at 4.2% of all HTML requests; PPC Land coverage
  • February 2026 - Ahrefs finds AI Overviews now correlate with 58% reduction in click-through rates for top-ranking pages
  • March 2026 - Cloudflare and ETH Zurich document AI bots breaking CDN cache layer; PPC Land coverage
  • March 2026 - AI bots crawl retail sites 198x more per visit than Google; PPC Land coverage
  • April 2026 - HubSpot reports 27% year-over-year organic traffic decline for customers; PPC Land coverage
  • April 2026 - OpenAI's web crawlers found to have tripled after GPT-5; PPC Land coverage
  • June 3, 2026 - Himanshu Sharma publishes LinkedIn post arguing AI brand mentions carry no commercial value
  • June 5, 2026 - Cloudflare Radar data shows bots at 57.4% of web traffic versus 42.6% human; shared by Cyrus S. on LinkedIn

Summary

Who: Cyrus S. (former Head of Global SEO), Himanshu Sharma (digital analytics professional), Rand Fishkin (SparkToro co-founder), Loganix, and the broader SEO and digital marketing community responding to current Cloudflare web traffic data.

What: Cloudflare Radar data published today shows automated bots now account for 57.4% of all HTML web traffic globally, with human visitors at 42.6%. Within that bot majority, AI training crawlers represent 50.6% of total traffic, while search crawlers account for just 10.7%. Separately, Himanshu Sharma has argued that AI brand mentions carry no measurable commercial value for businesses simultaneously experiencing organic traffic losses and declining leads.

When: The Cloudflare data was shared on June 5, 2026. The broader trajectory of rising bot traffic and falling organic traffic for publishers has built throughout 2025 and accelerated into 2026.

Where: The data originates from Cloudflare Radar, which measures traffic across the company's global network infrastructure protecting and proxying millions of websites. The professional discussion is taking place across LinkedIn, with the marketing and SEO community as the primary audience.

Why: The figures matter because they capture a structural shift in who and what is consuming web content. The majority of server requests are now generated by automated systems - most of them training AI models rather than directing users to websites. Combined with evidence that organic click-through rates have fallen sharply since Google AI Overviews expanded globally, and that brand mentions in AI responses do not appear to compensate commercially for the lost clicks, the data raises pointed questions about the economics of content publishing and the value propositions being offered to site owners by the AI industry.