Three traffic studies in four days found AI assistants send a sliver of the visits and sales Google does, which is the strongest public case yet against selling AI visibility on clicks.

Google added two sections to the page: one for the robots meta tag, data-nosnippet and X-Robots-Tag, and one for the Search generative AI control, which it describes as managing a site's inclusion in "AI Overviews and AI Mode" from Search Console. Nothing new was launched; the switch has been live for all sites since 31 August. What changed is that the opt-out now sits next to robots.txt on Google's most-read control page, where web agencies and site builders will find it.

Illyes said robots.txt rules apply to search crawlers, AI training crawlers and other well-behaved bots, that it is an access-preference protocol, and that it "doesn't dictate what a system does with the data" once crawled. To block one AI crawler, a site has to name its exact user-agent token. Nothing about Google-Extended, AI Overviews or AI Mode was said.

ChatGPT's newest entry is still locked chats (20 October). Perplexity's changelog is unchanged and marks no current Sonar model as deprecated. Anthropic's newest post (21 October) is a trustee appointment. Bing's webmaster blog still ends at 13 October. Cloudflare's 21 October posts cover a firewall feature, a PHP exploit and storefronts for shopping agents. Google's status dashboard shows no October ranking update despite the volatility reported on 20 and 21 October.

Marie Haynes, "GEO is a distraction" (21 October). Method in one clause: 12 months of GA4 data from 100 sites of varying size, over 145 million sessions, 55 of which tracked conversions; how AI sessions were identified, the date range and the site types are not stated. ChatGPT sessions were 0.62% of Google organic (0.73% with all assistants). ChatGPT visitors converted at 13.8% against 3.02% for Google organic, but Google organic drove 96.5% of conversions across the channels analysed: 964 conversions against 36 from all assistants combined. She calls the dataset small and says ChatGPT referrals plateaued around August 2025. The post does not address ChatGPT's October tracking bug, when visits were filed as direct.

SEO Works, AI referral traffic benchmark (undated, data September 2025 to August 2026). Method: 95 GA4 properties, AI visits matched by referring domain, 63 sites used for conversion comparison. 2,825,594 organic sessions against 72,095 AI sessions (39 times more organic). AI visits converted at 7.18% against 6.08% organic (1.18 times pooled), but worse than organic on 23 of 63 sites, and at 0.62 times organic on B2B product sites. The dataset is public.

Kevin Indig, Growth Memo, "Does AI drive more direct traffic?" (19 October). Method: GA4 direct traffic for 12 mostly B2B sites, August to September 2026, plus 632 lagged comparisons of weekly AI visibility against direct and branded search for four brands. Direct sessions carrying an identifiable AI trace were about 0.003% of direct traffic. One brand showed a 0.72 correlation between AI share of voice and direct visits three weeks later, but shifted (mismatched) timelines often fitted as well or better, so he found no consistent link. His conclusion is that hidden AI traffic could not be detected without asking customers how they found you.