Insights
What changed in search and AI answers, and what to check
Short, dated notes on changes that affect how pages are crawled, indexed and cited. Every one lists its sources.
Gemini and AI Mode user agents: a log test found “Google” and “Google-Lens”
Gemini and AI Mode user agents: one log test found Gemini fetches pages as “Google” and AI Mode as “Google-Lens”, HTML only. What to check on your site.
What is __usage_ai-training? The shared robots.txt AI training opt-out
__usage_ai-training is a shared robots.txt token that Google, Bing and OpenAI say they will honor for AI training preferences. What it does and what to check.
Google user-generated spam guidance: noindex shared pages by default
Google’s user-generated spam guidance now covers tools that publish shared pages, generated content and web apps: noindex link-shared pages by default and watch Search Console.
SISTRIX AI citation study: 96,418 URLs show which page features matter
The SISTRIX AI citation study compared 96,418 URLs cited or skipped by ChatGPT, AI Mode and AI Overviews. Structure, Article schema and dates stood out; llms.txt did not.
DeepSeek opened its search to AI agents. Can it read and date your pages?
DeepSeek launched Search and Fetch APIs on its in-house search engine on October 17, 2026. An early test found date filters drop undated pages and the fetcher skips JavaScript. What to check on your pages.
Cloudflare will grade bots on obeying robots.txt. Is your file saying what you mean?
Cloudflare is building trust bands that track whether Verified bots honor robots.txt and HTTP 429, and its pilot with OpenAI filters dead URLs and duplicates out of discovery. What to check on your site.
ChatGPT traffic didn’t disappear after GPT-6. Much of it is landing in Direct
Since GPT-6 shipped on October 7–8, 2026, many ChatGPT links arrive without a UTM tag or referrer, so analytics files them as Direct. The evidence so far and what to check this week.
Google’s UGC Fresh Data Program: a fast lane for forum posts that starts with your JSON-LD
Google documented the UGC Fresh Data Program on October 8, 2026. Who qualifies, the JSON-LD and freshness requirements, and what any forum or community site can check today.
Google now counts titles and headings as main content: what the October 2026 helpful-content update means on the page
On October 1, 2026 Google rewrote its helpful-content documentation around main content, listed page titles and headings as part of it, and set out the four attributes its quality raters use. What changed and what to check.
SEO didn’t die, it got stricter: search then and now, and the four gates to an AI citation
What changed when the first reader of your page became a model, what did not, and the four gates every page must clear before an answer engine can cite it. Two free graphics included.
Meta’s Muse agent books, buys and calls businesses. Here is how to be ready for it
Meta’s Muse AI agent reaches businesses three ways: through a cloud browser on your website, by phoning you, and through Meta AI Connectors. What each door needs, and what to fix first.
Cloudflare’s “Disallow AI Training” and Google’s AI opt-out: what changed for GEO in September 2026
Cloudflare now writes AI-training opt-outs into robots.txt without blocking Googlebot, and Search Console now reports and controls AI Overviews and AI Mode. What each change does, and what to check.