Pages the AI tools cite are four times as likely to open with the answer, and Google is putting ads into AI Mode follow-up questions.

Google is bringing ads in AI Mode to Australia, Canada, India, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore, with Brazil, Mexico and the UK planned for Q4, and is "launching multiturn ads in the U.S." so ads can appear in follow-up questions. Google says click-through rates "are already similar between sponsored and organic product list items within AI Mode" (internal US data, 20 August to 3 September 2026). The same day, Alphabet reported Search revenue up 12% year on year to $63.35 billion, flat on Q2, and its chief business officer named lower AI Mode monetization as one of three reasons growth slowed; no AI Mode usage figures were given.

ChatGPT's newest entry is still 27 October (iCloud Mail and Calendar). Google's changelog newest entry is still the 26 October spam guide. Perplexity's changelog shows only Agent API housekeeping, with no current Sonar model deprecated. Anthropic's newest post (28 October) is about its AI for Science program, nothing on search. Bing's newest webmaster post is still 13 October, and Cloudflare's newest posts are a security contest and a Q3 outage review.

Method in one clause: 193,507 answers (ChatGPT 67,543, AI Mode 63,033, AI Overviews 62,931) to 4,167 questions that named no company, for 78 of the agency's clients across nine business models (real estate, legal, home services, consumer products and B2B industrial are named; coaches are not), eight weekly snapshots from 27 August to 21 October 2026, collected with Peec AI, with AI raters blind-scoring 195 page pairs. In the blind round, 20% of cited pages opened with a direct answer against 5% of top-10 Google pages the AI never retrieved; against pages the AI retrieved but did not cite, it was 23% against 14%. Adjusted for domain rating and page type, odds of citation were higher with a year in the title (more than double), a date in the last 12 months (88%), tables (69%, 88% in ChatGPT), first-hand experience (45%), prices (28%) and schema markup (13%), and lower with an author byline (28% lower) and a call to action (12% lower). Engines differ: Reddit appeared in 16% of AI Overview answers but 1% of ChatGPT's, and YouTube in 23% of AI Overview answers and almost none of ChatGPT's.

The author says plainly: "These are associations." Search Engine Journal adds that the study "cannot show if putting in a direct answer for a question will change the number of times a web-page is cited", and that 23% of cited pages did not answer the question in the part raters read. The author's agency sells AI search optimisation, the Google data is one snapshot, and he cannot explain the byline and call-to-action results.