This week in search November 6, 2026

Adam Fry, who leads ChatGPT’s consumer product, told Alex Kantrowitz of Big Technology that ChatGPT’s volume of search-like queries is roughly one fifth of Google’s. He counts every question that sends ChatGPT to search the web. He also said weekly users grew from one billion to 1.2 billion in two months. Search Engine Journal notes he did not say which Google figure he compared against.

Microsoft Clarity added an Agent Activity dashboard that shows which AI agents visit a site and what they do there, from searches to bookings and checkout. It covers agents from Google, OpenAI and Anthropic, and can simulate agents attempting tasks. Microsoft’s post says weekly sessions from AI agents nearly doubled in three weeks to almost 2 million. The AI agent readiness guide covers measuring agents in server logs.

Cloudflare listed 14 problems AI agents create for websites and announced a Trusted Agent Delegation coalition to study them and bring the well-defined ones to the IETF as needed. A day earlier it began listing personal AI agents in its bot analytics and lets a site allow each one, block it on pages with ads, or block it outright. The AI crawler guide covers blocking at the CDN.

Robby Stein of Google said AI Mode has direct links to book or get quotes for home services in the US, with Angi, Thumbtack, Housecall Pro and Jobber as partners. He said it compared local pricing and checked reviews during his plumbing emergency. Search Engine Journal reports Jobber charges no lead fee for AI Mode requests from paid subscribers, and Housecall Pro promises the same once its integration goes live.

A LinkedIn post tagged for Search Central Live in Bengaluru quoted unnamed Google staff as anticipating click and query data for AI Overviews and AI Mode in Search Console. Gary Illyes replied that the team wants to give more data and has nothing to announce yet. Search Engine Journal reports the UK Competition and Markets Authority lists clicks among the AI metrics Google must provide by December 3.

Saul Westbrook found millions of AI-generated Facebook pages under facebook.com/fb-answers, each summarizing Facebook posts, and estimated from Semrush data that over 39 million rank in the US. He wrote that they survived the September spam update. Search Engine Roundtable counts about 6 million in Google’s index, and quotes Glenn Gabe calling it massive scaled content abuse.

The VLC media player site returned a 418 status code, which Google treats like most other 4xx errors. Gary Illyes said that because VLC is arguably one of the most important pages on the internet, Search temporarily protected the URL from falling out of the index. The result showed no snippet or favicon, only the link and the last good title. He expects it to sort itself out.

SE Ranking tracked 10,000 keywords and found the share with unrelated Google results jumped from 0.09 percent to 57.3 percent overnight on September 17, reaching 67.2 percent a week later. In 84 percent of cases the results were about one word of the query, and none started in the top five. SE Ranking says the cause may be Google trying to stop bots that scrape its results at scale.

Google rewrote the Crawled, currently not indexed entry in its Search Console help. It now says Google doesn’t index every page on the web and that the message is common and doesn’t necessarily point to an issue. For important pages it points to the URL Inspection tool, and to a form for suspected site-wide indexing bugs. Barry Schwartz pointed out that the new text makes no mention of quality issues.