When an AI gives you the answer upfront, do you still click through to where it came from?
Do users click links within AI summaries or end sessions instead?
This explores what people actually do when an AI summary appears at the top of a search: whether they click through to the sources behind it or take the answer and leave.
This explores whether AI summaries act as a doorway to sources or as an endpoint. The short answer from the corpus: mostly an endpoint. Pew's browsing-panel analysis of nearly 69,000 real Google searches found people clicked a search result 8% of the time when an AI summary appeared, versus 15% when there was none. Sessions were also 10 percentage points more likely to end with no click at all Do AI summaries on Google reduce clicks to actual websites?. Industry data points the same way. Ahrefs found the top organic result loses about 58% of its clickthrough rate on queries that show an AI Overview How much are AI Overviews actually reducing organic search clicks?. A natural experiment comparing Wikipedia's language editions found English Wikipedia lost roughly 5% of its search referrals once AI Overviews launched in English, with German and French serving as controls Does AI search summaries divert traffic away from Wikipedia?. One caveat: these notes measure clicks on ordinary search results. None of them isolates clicks on the citation links inside the summary itself, so the corpus can't say how often those get used.
Eye-tracking work shows why. The summary takes over the spot on the page where people look first. In an RMIT and Microsoft study, searchers fixated on AI Overviews far longer, and the share of attention going to the first-ranked result fell from 31% to 9%. Trust stayed equally high for both Where do searchers look when AI Overviews appear?. People aren't skipping the links because they distrust them. Their attention is already spent by the time they'd reach them.
The less obvious finding: better summaries make this worse. In Nextdoor experiments, LLM-written notification summaries were measurably more informative and got fewer clicks, because a good summary removes the reason to open anything Does better summary writing actually increase user engagement?. Summary quality and click-through can pull in opposite directions. The e-commerce work on ReLSum shows the reverse can be designed for. When summaries are trained against what users actually do next, rather than for fluent prose, engagement goes up Can reinforcement learning align summarization with ranking goals?. So whether a summary ends the session depends partly on what it was built to optimize.
Two threads complicate the simple story that AI answers replace the web. A cross-surface panel study found people more often open AI assistants after searching and browsing than before, by about 20 percentage points. That suggests many people still go to the web first and use assistants to finish the job Do people use AI assistants before or after searching?. Skipping the click may also have a learning cost. Across seven experiments with more than 10,000 people, those who learned from ChatGPT summaries reported learning less and felt less ownership of what they knew. The advice they then wrote was rated sparser than advice from people who had searched the web Does learning from AI summaries produce shallower knowledge than web search?. So the ended session isn't only a traffic problem for publishers. It may also mean shallower understanding for the reader who stopped at the summary.
Sources 8 notes
Pew's analysis of 68,879 Google searches found users clicked search result links 8% of the time when an AI summary appeared, versus 15% without one. Sessions were also 10 percentage points more likely to end without any clicks.
Ahrefs' December 2025 study found position-one CTR dropped to 0.016 for AI Overview keywords versus 0.039 for non-AI keywords, suggesting AI Overviews capture roughly 58% of clicks that would otherwise reach organic results.
A natural experiment using Wikipedia's language editions found that Google's AI Overviews lowered search referrals to English Wikipedia by 5.45% versus German and 4.82% versus French controls. The effect appears driven by answer-layer intermediation that satisfies user queries before clicks reach the source.
Eye-tracking data shows AI Overviews receive significantly longer fixation times, reducing attention to the first-ranked result from 31% to 9%. Trust ratings between AI Overviews and ranked results remained equally high despite this attention shift.
Nextdoor experiments showed LLM-generated summaries were objectively more informative but decreased click-through rates. Users had no reason to open notifications when the summary already satisfied their information need, demonstrating how optimizing for informativeness can backfire on engagement metrics.
Show all 8 sources
ReLSum trains summarizers using downstream relevance scores as RL rewards, producing dense, attribute-focused summaries instead of fluent prose. This alignment to the actual ranking metric improves recall, NDCG, and user engagement in production e-commerce search.
A cross-surface panel study found assistant sessions come after search and browsing 20.6 percentage points more often than before, reversing the "answer engine" narrative. Assistant-only sessions are also more common than search-only sessions within the same users.
Seven randomized experiments (n=10,426) show people who learned via ChatGPT reported less learning, felt less ownership of knowledge, and produced advice that independent raters found sparser and less informative than advice from web search users.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Google users are less likely to click on links when an AI summary appears in the results
- How AI Is Changing Search Behaviors
- Update: AI Overviews Reduce Clicks by 58%
- Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia
- An Eye Tracking Study: Are AI Overviews Changing Search Behavior?
- Emerging uses of AI chatbots for news and what it means for journalism (Digital News Report 2026)
- The New Shape of Search: How Conversational AI Recomposes Information Seeking
- News Source Citing Patterns in AI Search Systems