Google AI Overviews: More Links, But Not All Go Offsite

The Evolving Landscape of Google AI Overview Links

Google’s AI Overviews have undergone significant structural changes, introducing a variety of link types that behave differently depending on their placement and function. These elements include inline links adjacent to answer text, article suggestions at the conclusion of responses, previews of public discussions, and follow-up prompts that initiate conversations in AI Mode. Understanding how these components interact is critical for site owners, as they impact traffic attribution and user engagement metrics in distinct ways.

In Search Console, the classification of user interaction depends entirely on the destination of the click. A tap on a link directing users to an external website registers as both a click and an impression. Conversely, interactions that trigger a new internal Google query are excluded from both click and impression counts. This distinction creates ambiguity when analyzing performance data, particularly as Google experiments with more interactive features within its generative AI results.

Recent Observations and Industry Speculation

Recent developments have highlighted a surge in external linking within AI-generated answers. On September 21, Malte Landwehr of Peec AI published data indicating a sharp increase in this trend. The chart titled “External links in AI Overviews rise to 26%” illustrates the daily average percentage of AI Overviews containing external links within the answer text from August 25 to September 20. The data reveals a near-zero baseline until September 11, followed by a rapid ascent to approximately 26.2% by month’s end.

Landwehr noted the dramatic shift, stating: > “Google massively increased the number of external links within AI Overview answers.”

While the data suggests a clear upward trajectory, the methodology lacks specific details regarding sample size, geographic markets, or device types. Landwehr speculated that this shift might be part of a broader strategy to condition users to engage with links, potentially steering them toward AI Mode over time. He observed that some links now redirect users into conversational interfaces rather than traditional web pages.

Similarly, SEO expert Gagan Ghotra shared an 18-second video demonstrating this behavior. In the recording, clicking underlined text within an AI Overview does not open a webpage but instead launches a new query in AI Mode. Ghotra described this as another tactical move by Google to keep users within its ecosystem. While the feature has been available on mobile since January via the follow-up question prompt, Ghotra’s observation confirms that similar anchor-style links are now appearing directly within the answer text itself.

Google’s Official Stance on Link Integration

Google has progressively refined the display of links in AI Overviews, emphasizing enhanced user experience and content discovery. Robby Stein, VP of Product at Google Search, outlined several key updates throughout 2025 and early 2026. In August 2025, Stein announced plans to increase inline links and introduce link carousels on desktop devices. By late January, he highlighted the integration of follow-up questions, describing it as: > “one fluid experience with prominent links to continue exploring.”

Further refinements were detailed in February, where Stein explained that hovering over links would reveal descriptive icons and pop-ups, making it easier to navigate to high-quality content across the web. He stated that testing showed the new interface was: > “more engaging, making it easier to get to great content across the web.”

By May, Hema Budaraju, Vice President of Product Management for Search, provided a comprehensive overview of five major updates. These included labeled subscription links, previews of social media discussions, and additional inline links paired with relevant text. Budaraju emphasized that early testing indicated users were significantly more likely to engage with links identified as their personal subscriptions, though no specific metrics were released. She also confirmed that hover previews on desktop would display site names or page titles to aid decision-making.

Most recently, in June, Google acknowledged increasing the volume of inline links and continuing to experiment with diverse link designs. However, official communications have consistently omitted precise figures regarding market rollout timelines or the exact proportion of queries affected by these changes.

Decoding Search Console Metrics

Navigating the analytics landscape for AI Overviews requires careful interpretation of Search Console data. The platform treats different link types with varying levels of attribution. When a user clicks a link to an external page within an AI Overview, it is recorded as a standard click and impression. However, if the link functions as a query refinement,triggering a new search,the interaction is ignored for both metrics.

The generative AI performance report, fully rolled out globally on August 31, provides insights into impressions but excludes click data. This report aggregates impressions by property, meaning multiple links to the same domain within a single response count as one impression. Individual page-level data is available, but it cannot be used to calculate a traditional click-through rate (CTR) because the total Web search performance report includes clicks from all sources, not just AI features.

This limitation creates a blind spot for publishers. A site may see a spike in AI impressions without knowing if those impressions translated into actual visits. Furthermore, if Google classifies certain interactive elements, such as the anchor links demonstrated by Ghotra, as query refinements, they will not generate any trackable clicks or impressions for external sites. Until Google clarifies how these specific interactions are categorized, site owners cannot definitively attribute traffic losses or gains to these emerging link formats.

Implications for Content Strategy

The divergence between impression data and click tracking poses challenges for measuring the true impact of AI Overviews. For news and content publishers, an increase in AI impressions does not guarantee an increase in referral traffic. Since clicks from AI features are bundled with general search clicks, isolating the performance of specific AI-driven links remains difficult.

Third-party trackers like Peec AI offer valuable indicators of UI changes, such as the rising prevalence of external links. However, these metrics reflect display frequency rather than user engagement or traffic outcomes. A higher percentage of linked answers indicates a change in Google’s presentation layer, not necessarily a change in user behavior or site visibility.

As Google continues to integrate more interactive elements, including potential redirects to AI Mode, the line between traditional search results and conversational interfaces blurs. Publishers must remain vigilant about these shifts, recognizing that current reporting tools may not fully capture the nuances of user interaction. Future updates to Search Console are expected to provide more granular metrics, but until then, interpreting AI Overview performance requires a holistic view of both impression trends and overall search traffic patterns.