Google AI Mode: Study Reveals Negative Impact on User Experience and Publisher Traffic

| Editorial Team On 23 Sep, 2026

Google AI Mode Drives Users Away and Cuts Publisher Traffic, Study Finds

New research tracking over 1,000 users contradicts Google's own claims that AI Search delights users and sends more traffic to the open web.

A landmark study published in August 2026 has delivered a damaging verdict on Google's AI Search features. Researchers from the University of Pennsylvania and Northeastern University found that Google's AI Mode reduces clicks to publisher websites and creates such a poor user experience that it pushes users toward rival search engines.

The draft paper titled AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence directly challenges statements made by Google leadership — including CEO Sundar Pichai — who has repeatedly claimed that AI Mode is sending more users to the open web than ever before. For publishers, SEO professionals and digital marketers already struggling with declining organic traffic, the findings carry serious implications for how they plan and measure their search visibility strategies.

What the Research Found

The study enrolled 1,444 participants and tracked the search behavior of 1,100 of them over a seven-day field experiment. Of those tracked users, 956 completed a post-experiment survey. Researchers used a browser extension to record search behavior, collect browsing history and capture snapshots of Google search results pages.

Participants were randomly assigned to one of three search conditions:

The first group was forced into regular Google Search with AI Overviews hidden and AI Mode redirected
The second group experienced natural Google behavior, including AI Overviews when Google chose to display them
The third group had all searches redirected into AI Mode

The randomized design allowed researchers to isolate the precise effect of each AI feature on user behavior and satisfaction. The methodology is notable for its real-world application rather than relying on self-reported preferences or lab simulations — a distinction that lends the findings considerably more weight than typical survey-based research.

A Notable Technical Obstacle

Researchers flagged one significant technical obstacle during the experiment. Google updated the HTML structure of AI Overviews mid-study, which broke the browser extension's ability to hide them. As a result, only 51.1% of AI Overviews were successfully hidden across the experiment's duration. The AI Mode routing performed better, with 94.7% of searches in that group successfully redirected.

This mid-study disruption is worth noting not just as a methodological caveat, but as an illustration of how rapidly Google alters its search infrastructure — a challenge familiar to anyone working in SEO or content strategy. Understanding the risks and operational challenges AI poses for business has never been more relevant, and this study adds empirical weight to concerns that have largely been anecdotal until now.

AI Mode Reduces Clicks and Drives Users to Competitors

The study's most consequential finding is that AI Mode substantially reduced the likelihood of users clicking through to external websites. When AI Overviews were removed from search results, clicks to publisher websites increased with no measurable decline in user satisfaction. This suggests that AI Overviews function as a substitute for visiting source websites rather than a gateway to them.

Fewer Searches, Not More

AI Mode also had an unexpected effect on search activity overall. Researchers had anticipated that AI Mode's richer answers might encourage users to search more. The opposite occurred. Users assigned to AI Mode conducted fewer search sessions per day. Users in the no-AI condition showed no change in their daily search activity in either direction.

The Competitive Fallout

Perhaps most striking is the competitive dimension of the findings. The study found that assignment to AI Mode significantly increased the fraction of users who searched on a competing engine — such as Bing, DuckDuckGo or Yahoo. It also increased stated intention to switch to Bing permanently. In a market Google has dominated for decades, this represents a meaningful signal of user dissatisfaction that Google's public communications have done nothing to acknowledge.

The researchers also noted that AI Mode's negative effects were amplified among heavy users of Google Search. Those who relied on Google most frequently reduced their search activity more sharply when exposed to AI Mode than lighter users did. For brands and publishers whose audiences skew toward frequent, research-oriented searchers, this finding has direct implications for expected referral volumes.

What This Means for Publishers and Marketers

Google's Public Position Is Under Scrutiny

Google's leadership has been consistent and vocal in its defense of AI Search. Liz Reid and Sundar Pichai have both stated publicly that users are delighted with AI Mode and that engagement with it is strong. Pichai specifically claimed the feature is sending more traffic to the open web than at any previous point.

The research findings sit in direct opposition to those claims across multiple dimensions. AI Mode worsened how users perceived Google Search overall. It reduced trust in search results. It lowered users' sense that search was useful or personally relevant. And it reduced their satisfaction with the experience.

Removing AI features produced no detectable negative change in how users perceived search. That finding is particularly significant because it undermines the argument that AI Mode is necessary for a quality search experience.

One area where AI Mode showed no clear difference in user perception was news-related searches. The study found no statistically significant variation in how users responded to AI Mode versus regular search when their queries were news-oriented.

The question being raised across the SEO and publishing community is no longer simply whether AI Mode reduces traffic. It is whether Google's public statements about AI Mode bear any meaningful relationship to what its own data shows. For further context on how search experience is evolving and what that means for visibility strategy, the broader shift toward search experience optimization is an important framework for understanding where search is heading.

Three Strategic Responses for Search Professionals

For professionals managing search strategies and content investments, the study surfaces clear and pressing implications.

First, declining referral traffic from Google is not a measurement error or a temporary fluctuation. It reflects a structural feature of how AI Mode is designed to keep users within Google's ecosystem. Diversifying traffic sources beyond Google Search is no longer optional strategy — it is risk management. Publishers and marketers who have not yet mapped out alternative acquisition channels should treat this research as an urgent prompt to do so. Practical steps to increase website traffic beyond search engine referrals are worth revisiting in light of these findings.

Second, tracking traffic in analytics tools requires closer attention to the direct and other bucket, where AI-sourced visits often go unattributed. Understanding the true source of referrals is essential before making budget or content decisions based on incomplete data. If your analytics show a growing share of direct traffic alongside declining organic, AI Mode may be the explanation rather than increased brand awareness.

Third, the research reinforces the value of building direct audience relationships through email newsletters, owned communities and branded content. As AI search continues to reduce the reliability of organic click-through, the publishers and brands with direct connections to their audiences will be most insulated from continued losses. This is not a new idea — but this study provides the clearest experimental evidence yet that it is the correct one.

The full draft paper, AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence, is available via SSRN for those who want to examine the methodology and findings in detail.