Google has officially addressed concerns regarding the functionality of its Discover AI summary feature, confirming that a recent issue preventing users from clicking through to source websites was the result of a software bug within a temporary experiment. The confirmation came directly from Rajan Patel, Vice President of Engineering for Search at Google, who clarified that the company has since deactivated the test and restored the expected user experience. This development follows a period of uncertainty among digital publishers and SEO professionals who had observed unexpected behavior in the way Google’s AI-generated content interacted with external citations.
The Chronology of the Discovery
The issue first gained traction on October 23, 2024, when Google Discover researcher Damien Andell identified a technical anomaly within the platform’s interface. According to Andell’s observations, AI-generated summaries appearing in the Discover feed—a personalized content stream for mobile users—had suddenly lost their ability to navigate users to the original source material.
Under normal operating conditions, tapping on the "Sources" section of an AI-generated summary card triggers an expansion, allowing users to view a curated feed of articles from the cited publications. However, Andell discovered a specific internal flag within the Google app labeled "disable_aio_sources_button_tap." When this flag was active, the interactive elements intended to drive traffic to publishers were rendered unresponsive.
The report caught the attention of prominent industry analysts, including Lily Ray, who highlighted the potential ramifications for publishers who rely on Google Discover as a significant traffic driver. The discussion quickly moved to X (formerly Twitter), creating a centralized forum for industry experts to troubleshoot the behavior. Following this public discourse, Rajan Patel intervened to provide clarity, stating that the disabled taps were not a strategic policy change but rather an unintended consequence of an internal experiment. By October 27, Andell reported that the specific "disable" experiment had been removed and replaced with a new internal test identified as "aio tap falls through to hero."
Understanding the Role of AI in Google Discover
To understand why this bug caused such significant concern, one must examine the evolving nature of the Google Discover ecosystem. In recent years, Google has integrated generative AI more deeply into its user interface, aiming to provide concise, summarized information directly within the feed. These "AI Overviews" or summaries are designed to synthesize content from multiple web sources, offering users a quick digest of news or topics of interest.
For publishers, the value of appearing in these summaries is twofold. First, they gain visibility through the summary itself, which contributes to brand authority. Second, and perhaps more importantly, the "source link" functionality serves as a primary funnel for organic traffic. When Google inhibits the ability of a user to navigate from a summary to a full article, it effectively severs the traffic loop, turning a potential acquisition channel into a closed-loop information repository.
Search Console and the Metrics of Visibility
The incident underscores the growing complexity of measuring traffic in an AI-augmented search landscape. Google’s Search Console recently introduced specific reporting tools for "Discover Generative AI," which allow webmasters to track how often their content is surfaced within these AI summaries.
According to official Google support documentation, the primary metric provided in these reports is "Impressions." An impression is logged when a user views a link to a website through the generative AI feature in Discover. Crucially, this metric does not differentiate between a user who merely views the summary and a user who clicks through to the website.
The bug reported by Andell highlights the limitations of current tracking metrics. If the "source tap" functionality is disabled—either by design or by accident—publishers might see their impression data remain stable while their actual referral traffic drops significantly. This creates a data discrepancy that can lead to confusion regarding the effectiveness of a site’s content strategy. As Google continues to experiment with these features, the industry faces an ongoing challenge in reconciling "visibility" with "engagement."
The Implications of Internal Experiments
The fact that this issue was linked to an internal experiment raises broader questions about how Google tests features that directly impact the publisher ecosystem. Large-scale technology platforms often run "A/B tests" or "flagged experiments" to determine how interface changes affect user retention and interaction time. In this instance, the experiment appears to have been an attempt to test a "closed" experience, where the AI summary acts as the final destination rather than a gateway.
While Google maintains that this was a technical bug rather than a policy shift, the incident serves as a reminder of the volatility inherent in search and discovery platforms. For SEO professionals and content creators, the incident highlights the necessity of constant monitoring. Because many of these experiments are rolled out to small segments of the global user base, a "bug" for one user may be a fully functional feature for another, making it difficult to maintain a standardized view of performance metrics.
Industry Response and Future Outlook
The reaction from the SEO community was one of relief, tempered by a renewed sense of vigilance. When news broke that the experiment had been deactivated, many professionals viewed it as a confirmation that Google remains committed to the current referral model, at least for the time being. However, the subsequent replacement of the bugged experiment with a new test ("aio tap falls through to hero") suggests that Google is still actively iterating on the user interface for AI summaries.
The "hero" terminology often refers to the primary, high-visibility element on a page. If the new experiment changes how taps function, it could mean that clicking an AI summary might now lead directly to the primary article or a landing page rather than a list of sources. Publishers will need to keep a close eye on their referral traffic logs to see if these ongoing experiments impact their click-through rates (CTR).
Best Practices for Publishers in an AI-First World
In light of this event, publishers are advised to take several precautionary steps to ensure their content remains visible and accessible:
- Monitor Referral Traffic Patterns: Publishers should cross-reference Search Console data with their own internal analytics. If there is a sudden, unexplained drop in Discover traffic that does not align with a decrease in impressions, it may be worth investigating if the source links are still functioning correctly.
- Engage with Technical Communities: The discovery of this bug was entirely community-led. Engaging with forums, professional networks, and industry researchers allows publishers to identify and respond to platform changes faster than they might through official channels alone.
- Diversify Traffic Sources: While Google Discover is a massive driver of traffic, the volatility of AI experiments reinforces the need for a diversified traffic strategy. Relying solely on platform-specific features that are subject to constant testing can expose a business to unnecessary risk.
- Prioritize Quality and Context: Regardless of how Google chooses to display content, the underlying quality of the journalism or information remains the most consistent factor. If content is deemed valuable, the platform is incentivized to provide a path for users to reach it, as it improves the overall quality of the Google search experience.
Conclusion
The resolution of the "disabled source tap" issue provides a moment of clarity for the publishing industry, confirming that the current friction was an experimental error rather than a deliberate move toward a walled-garden architecture. However, the event serves as a microcosm of the current state of search: a highly dynamic environment where features are in a constant state of flux.
As Google continues to refine its generative AI capabilities, the line between "search" and "discovery" will continue to blur. Publishers must remain agile, utilizing both official reporting tools and community-sourced data to navigate the changing landscape. For now, the "source tap" remains a vital bridge between the AI summary and the source, and the industry can proceed with the assurance that Google is aware of the importance of maintaining that connection. As new experiments continue to roll out, the priority for site owners remains clear: continue to produce high-quality, relevant content while keeping a watchful eye on the technical conduits that bring that content to the world.
