Research Shows Google AI Mode Sends Less Clicks & Is a Poor User Experience

For years, we have been told that the evolution of search is about making information more accessible. But there is a fine line between providing a quick answer and creating a wall between the user and the source of that information. When a search engine stops being a gateway and starts acting as a destination, the entire ecosystem of the open web begins to shift.

Recent data suggests that this shift is not only hurting publishers but is actually frustrating the people using the tool. If the goal of AI in search was to improve the experience, the evidence suggests it might be doing the exact opposite.

The Gap Between Google's Claims and Reality

A draft paper published in August 2026, titled AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence, brings some uncomfortable truths to light. The research, conducted by teams from Northeastern University and the University of Pennsylvania, suggests that Google's AI search features are actively reducing the amount of traffic reaching publisher websites. This connects with Is Google Fixing B2B Marketing? when the same signal needs a clearer operating decision.

More concerning is the impact on the user. The study indicates that AI Mode creates a user experience so poor that it actually drives people to look for alternatives like Bing, Yahoo, or DuckDuckGo. This stands in direct opposition to the narrative pushed by Google, which claims that users are enamored with AI search and that it is driving more traffic to the web than ever before.

The core findings are straightforward: AI search features cut down on website traffic, lower user satisfaction, and keep users trapped within the Google ecosystem. Interestingly, when AI Overviews were removed, clicks to external sites increased without any negative impact on the user's overall search experience. The same pattern also shows up in How Travel Brands Can Earn AI Recommendations, where the practical question is how the signal becomes visible.

Expert Interpretation: This is a critical distinction. Google often frames AI Overviews as a "value add." However, if removing the feature increases clicks without hurting the user experience, the feature isn't adding value to the user; it is adding value to Google by keeping the user on their own page. For anyone relying on organic traffic, this confirms that the "zero click" trend is being accelerated by design, not by accident.

How the Study Was Conducted

To get these results, the researchers didn't rely on anecdotes. They ran a field experiment with 1,444 participants, tracking the actual search behavior of 1,100 of them over a week. To ensure the data was clean, they used a browser extension that captured snapshots of search result pages and recorded exactly where users clicked.

The participants were split into three different groups to see how they reacted to different environments:

Forced Regular Search: AI Overviews were hidden, and any AI Mode searches were sent back to standard results. Natural Behavior: The standard Google experience, where AI Overviews appeared whenever Google's algorithm decided to show them. Forced AI Mode: Searches were intentionally redirected into the AI Mode experience.

The researchers noted some technical hurdles, mentioning that while most AI Mode searches were successfully routed, the ability to hide AI Overviews in the "No AI" group declined over the course of the experiment. Despite these frictions, the data provided a clear look at how users behave when they are forced into an AI centric search flow versus when they are not.

Expert Interpretation: The use of a browser extension to capture real time behavior is far more reliable than a self reported survey. It removes the "recall bias" where users might say they like a feature but their actions show they avoid it. When you see a direct correlation between the presence of AI and a drop in clicks, you are looking at a behavioral fact, not an opinion.

AI Overviews as a Catalyst for Engine Switching

One of the most surprising findings is that AI Mode doesn't just keep users on Google; it actually pushes some of them away from Google entirely. The research shows that when users were forced into AI Mode, many found the experience so frustrating that they began using competitor search engines. A useful companion note is New Research Shows It’s Mostly Statistical Noise, because it looks at a nearby part of the same system.

The data indicates a significant increase in the fraction of users searching on Bing, DuckDuckGo, or Yahoo when assigned to AI Mode. The intention to switch engines became a measurable outcome of a poor AI experience.

Essentially, by trying to "trap" the user within a generated answer, Google may be creating enough friction to make the user seek a more traditional, link based experience elsewhere.

Expert Interpretation: This suggests a "breaking point" for user patience. There is a limit to how much synthesized information a user will accept before they realize they actually need a primary source. If Google continues to prioritize the AI answer over the source link, they risk alienating the very users they are trying to retain.

The Decline of the Click Through Rate

The data confirms what many publishers have suspected: AI search features act as a substitute for visiting a website. Users who were redirected to AI Mode were substantially less likely to click through to external sites.

Conversely, when AI Overviews were removed from the experience, clicks to external websites went up. This proves that AI Overviews are not "discovery tools" designed to help users find the best source; they are "answer engines" designed to prevent the user from needing to leave the search page.

Expert Interpretation: This is the "substitution effect." If a user can get a "good enough" answer from an AI summary, they won't click the link. The tradeoff here is accuracy and depth for convenience. As a site owner, this means your content is being used to train and fuel the AI that is now preventing users from actually visiting your site.

Reduced Search Frequency in AI Mode

The researchers initially hypothesized that AI Mode might actually encourage people to search more, perhaps because the AI makes complex queries easier. The results showed the exact opposite.

People assigned to AI Mode actually conducted fewer search sessions per day. Interestingly, those in the "No AI" group showed no change in their search frequency. This means that while removing AI didn't make people search less, adding AI actually discouraged them from searching as often.

Expert Interpretation: This is a counterintuitive result. Usually, a better product leads to more usage. If AI Mode is reducing the number of searches, it suggests that the experience is either unsatisfying or creates a level of friction that makes the user less likely to return to the search bar.

Impact on User Trust and Perception

The psychological impact of AI Mode was overwhelmingly negative. Users reported that AI Mode made them feel worse about Google Search in general. Specifically, the research noted a decrease in:

Overall trust in the search results. The perception that the search was useful or satisfying. The feeling that the results were personally relevant.

Crucially, removing the AI features did not make users perceive the search experience as worse. It simply returned them to a baseline of satisfaction that the AI was actively eroding.

Expert Interpretation: Trust is the primary currency of a search engine. If users feel the AI is hallucinating, oversimplifying, or hiding the real sources, the trust in the entire brand diminishes. The fact that "No AI" didn't lower satisfaction proves that the AI is the variable causing the decline.

The Nuance of News Related Searches

It is worth noting that the negative perceptions weren't universal across all types of queries. When it came to news related searches, the researchers didn't find a clear difference in how users perceived AI Mode compared to regular search.

This may be because news is inherently timely and source dependent, and users may already have different expectations for how news is delivered compared to general knowledge queries.

Expert Interpretation: This suggests that AI might be more "tolerable" in specific contexts, but it doesn't negate the overall trend. Even in news, the risk of the "zero click" remains, even if the user doesn't explicitly report a drop in satisfaction.

Why Heavy Users are the Most Affected

The study found that not all users reacted to AI Mode in the same way. The negative effects were most pronounced among "heavy users" of Google Search.

For those who use search engines frequently and intensively, AI Mode led to a much sharper decline in search activity than it did for casual users. Heavy users likely have a more refined understanding of how to find high quality information and are more sensitive to the friction or inaccuracies introduced by AI summaries.

Expert Interpretation: This is a dangerous trend for Google. Heavy users are the power users who provide the most data and the most consistent engagement. If the AI experience is alienating the most proficient users, Google is effectively degrading its most valuable user segment.

Questioning the Corporate Narrative

There is a stark contrast between this research and the public statements made by Google leadership. From Liz Reid to CEO Sundar Pichai, the company has consistently claimed that users are delighted with AI Mode and that it is driving more traffic to the open web.

The evidence from the University of Pennsylvania and Northeastern University suggests the opposite. The data indicates that AI Mode traps users on the search page and reduces the flow of traffic to the rest of the internet. This aligns with the lived experience of publishers who have seen their organic traffic plummet since the rollout of AI Overviews.

When the data from a randomized field experiment contradicts the claims of a CEO, it raises a fundamental question about the transparency of the AI rollout. If the user experience is objectively worse and the traffic to the web is decreasing, the narrative of "mutual benefit" for users and publishers appears to be a fabrication.

Expert Interpretation: For the digital strategist, this is a signal to diversify. If the primary gateway to the web is intentionally designing a system that reduces referrals, relying solely on search traffic is a high risk strategy. The focus must shift toward owned audiences, newsletters, and direct brand loyalty, as the "open web" is being enclosed by an AI wall.