People still click after reading an AI answer — they just take longer to do it. Search Engine Land highlighted Fractl research on November 7 showing how search sessions have changed from 2024 to 2026: the median time to click a search result went from 2.5 seconds to 6 seconds in two years. Fewer sessions end in a bounce, and more end in a re-query. The read from the research is that the clicks that survive AI answers carry more intent. The finding reframes the AI-search debate: the question is not only how many clicks are lost to AI answers, but what the remaining clicks are worth — and the evidence says they are worth more, because the person clicking has already read the answer and decided they want more.
The median time to click more than doubled in two years, from 2.5 seconds to 6 seconds. That is a large behavioural shift, and the explanation is straightforward: the AI answer now sits between the query and the click. The searcher reads the summary first, weighs it, and only then decides whether to click through. The click is no longer the first action after the query — it is the second, taken after the AI has had its say.
That delay is not indecision; it is evaluation. A searcher who clicks after six seconds has read the answer, found it insufficient or intriguing, and chosen to go deeper. A searcher who clicked after 2.5 seconds in 2024 was often clicking on the basis of a title and a snippet alone. The slower click is the more considered one, and the more considered click is the one likelier to convert.
The session shape has changed too: fewer sessions end in a bounce, and more end in a re-query. A bounce — a click followed by an immediate exit — is the signature of a searcher who found the wrong page. If fewer sessions end that way, it suggests the AI answer is doing some of the filtering: people who would have clicked the wrong result and bounced now get their answer from the summary, or re-query with a better question instead.
The rise in re-queries is the other side of the same coin. Search is becoming more conversational: the first query gets an AI answer, the searcher refines the question, and the session continues. That is a different journey from the old one-query-one-click pattern, and it rewards content that answers the follow-up question, not just the first one. Pages that serve the second and third query in a session will earn the clicks that the first query no longer gives away.
The measurement implication is direct: time to click, bounce rate and re-query rate are becoming as important as click-through rate. A page whose CTR fell but whose visitors now arrive after six seconds of reading an AI answer may be getting better traffic, not worse — visitors who chose the page deliberately rather than by default. Reporting CTR alone, without the session context, will mislead.
For content strategy, the lesson is to write for the considered click. The searcher who arrives after reading an AI summary has already seen the generic answer — they want the detail, the evidence, the original data, the thing the summary could not give them. Pages that offer depth beyond the summary will earn the clicks that survive. Pages that merely repeat the summary will not be clicked at all, because the summary already said it.
The six-second figure also reframes what a good snippet looks like. In a world where the searcher reads the AI answer before deciding to click, the snippet's job is no longer to win the click on its own — it is to promise something the answer did not provide. A title that merely restates the query will lose to the summary every time, because the summary already answered it. A title that promises depth, evidence, or a specific angle gives the searcher a reason to spend the six seconds and then click. The pages that win the considered click will be the ones whose titles and descriptions read as an invitation to go deeper, not as a repetition of what the AI already said.
Sources: Search Engine Land, November 7, 2026 (Fractl research).
Related reading on this site: geo-referral-data-misses-buying-decisions, Early-November Google Volatility Hits 9.4 Out of 10 — With No Update Confirmed, ChatGPT Ads Switch to Target Bidding on November 9 — Max Bid Retires January 1, Google AI Mode Now Books Home Services — Angi, Thumbtack, Jobber and Housecall Pro, Google AI Mode Now Shows Up to Four Ads Under a Single Answer, and Facebook Answers Drew an Estimated 40.6 Million Google Visits in October.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial