Insights · Field note · Growth and AI visibility
Why answer engines changed demand, in five sourced numbers
The shift from ranked links to generated answers is measurable. Here are the figures we use when a client asks whether this is real.
When an AI summary appears on a Google results page, users click an ordinary result on about 8 percent of visits, against 15 percent when it does not (Pew Research Center, 2025). Google AI Mode sessions end without a click about 93 percent of the time (Semrush, 2025). Visitors arriving from AI search converted at roughly 4.4 times the rate of organic visitors in a cross-property study (Semrush, 2025). Nearly 90 percent of pages cited by ChatGPT ranked at position 21 or worse on Google for the same query (Semrush, 2025). ChatGPT reported one billion weekly users in August 2026 (OpenAI).
In detail
Five numbers and what each means operationally.
Clicks fall when a summary appears
Pew’s 2025 analysis found that on results pages showing an AI summary, users clicked a traditional result on about 8 percent of visits, against about 15 percent when no summary appeared. Operationally: a page that depends on the click from a ranked result loses roughly half of that traffic on queries where Google generates a summary. The question becomes whether the summary names you.
AI Mode is a zero-click surface
Semrush measured a zero-click rate of about 93 percent for Google AI Mode across roughly 69 million U.S. desktop sessions between May and July 2025. In the same body of work, AI Overviews appeared on about 15.69 percent of queries measured in November 2025. Operationally: on the conversational surface, the answer is the whole experience. Being cited inside it is the only visibility there is.
Fewer visitors, better ones
Semrush reported that visitors arriving from AI search converted at about 4.4 times the rate of organic search visitors in its cross-property analysis. Ahrefs reported a 23 times conversion rate for AI-search visitors on its own site, which is a single-site observation and should be read as such. Operationally: measure qualified inquiries by source, not sessions. The AI-referred visitor has already been told who you are.
Citations do not follow rankings
Semrush found that almost 90 percent of the pages ChatGPT cited sat at position 21 or worse in Google for the same query. Operationally: ranking first does not earn the citation, and ranking nowhere does not prevent it. Retrieval selects the passage that answers the question from a source it can attribute, which is a different competition with different winners.
The audience is already there
OpenAI announced one billion weekly ChatGPT users in August 2026. Goodie’s 2026 study of 41 B2B brand sites attributed 62.6 percent of B2B AI referrals to ChatGPT and 18.5 percent to Claude between March and April 2026. Operationally: this is not an emerging channel to watch. It is where a large share of buyers already ask their first question, and the measurement program should treat it as such.
Sources
Primary references.
- Pew Research Center, July 2025. Analysis of click behavior on Google results pages with and without AI summaries; 900 U.S. adults, browsing data from March 2025.
- Semrush, 2025. AI Mode and AI Overviews study: about 69 million U.S. desktop sessions, May to July 2025; AI Overviews trigger rate measured November 2025; AI-search visitor conversion and ChatGPT citation position analysis.
- Ahrefs, July 2025. Freshness of AI-cited URLs versus organic results across 16.975 million citations; conversion rate of AI-search visitors on ahrefs.com (single site).
- Goodie, 2026. AI search traffic report across 41 B2B brand sites, March to April 2026.
- OpenAI, August 2026. Announcement of one billion weekly ChatGPT users.
By David Selakovic · Published 2026-10-16 · Revised 2026-10-16 · Reviewed every 180 days