Google Tests New Dive Deeper AI Feature in Discover Feed to Expand User Engagement and Traffic Referral Pathways

Google is actively testing a new user interface feature within its mobile Discover feed called the Dive deeper button, an addition designed to streamline how users explore complex topics and discover related web content. Announced late last week by Robby Stein, Vice President of Product for Google Search, the ongoing experiment represents the latest iteration of Google’s broader strategy to integrate generative artificial intelligence deeply into its content discovery ecosystems.

The initial rollout of the test focuses exclusively on video content, leveraging AI-generated summaries paired with curated source links to provide users with a comprehensive snapshot of trending subjects. While the company has not yet disclosed specific geographic regions, language parameters, or user-base percentages for the test, the initiative has already sparked significant interest among digital publishers, search engine optimization professionals, and industry analysts monitoring the shifting dynamics of referral traffic.

Mechanics of the Dive Deeper Experiment

The functionality of the new feature centers on bridging the gap between passive content consumption and active exploration. According to demonstrations shared by Google leadership, when a user encounters a qualifying video card within their Discover feed—such as a local news clip detailing a regional weather event—an AI-generated summary appears directly beneath the media. This initial summary is clearly marked with a standard disclaimer stating it was "Generated with AI, which can make mistakes."

Positioned immediately below this summary is the new Dive deeper button. Tapping this element opens a dedicated modal or expanded overview screen. This interface features a more detailed topic synopsis accompanied by a horizontal carousel of prominent source cards and links. These direct pathways direct users toward original reporting, alternative coverage, and verified community reactions.

A subsequent screen within the experimental layout, tentatively titled Explore more stories, groups shorter text summaries under granular sub-topic headlines. Each subsection is supplemented by direct attribution links and brief descriptions of the external coverage, creating a localized knowledge hub directly within the Discover environment. By organizing diverse perspectives and primary reporting into a single interactive layer, Google aims to keep users engaged while simultaneously routing traffic to web publishers whose content informs the AI-generated overviews.

Background and Context within Google Discover

This latest experiment does not exist in a vacuum; it builds upon a series of rapid deployments aimed at infusing Google Discover with conversational and generative AI capabilities. In October 2025, Google rolled out AI previews on trending topics within the Discover and sports feeds. Those initial previews were designed to expand upon user taps, revealing supplementary contextual information alongside foundational web links. That rollout was initially deployed across select key markets, including the United States, South Korea, and India, setting the standard for how information is packaged outside of traditional search results pages.

The integration of generative AI into Discover has also necessitated updates to webmaster analytics tools. Google Search Console recently introduced a dedicated generative AI performance report tailored specifically for Discover traffic. Formally rolled out globally on August 31, the reporting mechanism is designed to capture and display impressions for links appearing within generative features across the Discover feed.

However, webmasters and digital marketers have noted lingering transparency challenges regarding these analytics. Official Google documentation indicates that while the report aggregates data by page, country, and date, it relies primarily on impression counts as its core metric. Furthermore, the help documentation currently lacks a comprehensive itemization of precisely which experimental features qualify for tracking, and explicitly excludes tests operating within Search Labs. Because the new Dive deeper button is currently classified as an experiment, industry professionals remain uncertain whether traffic and impressions generated through this specific interface will be clearly isolated within Search Console or blended into broader AI feature metrics.

Chronology of AI Integration in Search and Discovery

To understand the trajectory of the Dive deeper feature, it is helpful to examine the timeline of Google’s recent interface transformations:

• Early 2024 to Mid-2025: Google scales its Generative Experience across core search results, laying the technical foundation for context-aware summaries across mobile and desktop platforms.
• October 2025: Google introduces AI-powered previews for trending topics within the Discover feed in select international markets, offering expanded contextual insights on key news items.
• August 31, 2025: Google deploys the Search Console generative AI performance report for Discover worldwide, attempting to provide publishers with visibility into AI-driven referral streams.
• September 2025: Official documentation updates outline the tracking parameters for Discover AI features, though distinctions regarding specific experimental modules remain ambiguous.
• Late 2025 / Early 2026: Robby Stein announces the experimental rollout of the Dive deeper button on X, beginning with video content and paving the way for multi-design iterations.

Industry Implications and Publisher Perspectives

For digital publishers, content creators, and SEO strategists, the introduction of the Dive deeper button carries a complex mix of potential benefits and operational concerns. On one hand, the inclusion of direct source cards and prominent attribution links within the expanded overview offers a formalized pathway for referral traffic. Unlike traditional zero-click search results that satisfy user queries entirely on the search engine results page, the Discover AI overview explicitly encourages users to venture deeper into original reporting and related coverage.

Conversely, the reliance on AI-generated summaries introduces questions regarding user retention and attribution fairness. If a user is satisfied by the concise breakdown provided within the Dive deeper interface, they may choose not to visit the underlying publisher sites, potentially altering traditional click-through rates. Additionally, the opacity surrounding how impressions and clicks from these experimental modules are recorded in Google Search Console complicates traffic forecasting and performance attribution for media organizations heavily reliant on organic referral streams from mobile feeds.

As digital media consumption increasingly shifts toward algorithmic feeds rather than intentional keyword queries, the mechanics of content discovery hold immense economic weight. Publishers must constantly adapt their optimization strategies to align with Google’s shifting presentation layers, ensuring their content is structured in a way that appeals both to human readers and the underlying large language models curating the summaries.

Future Outlook and Iterative Development

Google has emphasized that the current deployment of the Dive deeper button is strictly an initial test. In his public statements, Robby Stein noted that the product team will experiment with multiple visual designs and user-experience configurations over the coming weeks. This iterative approach allows Google to gather telemetry data on user engagement, tap-through rates, and interaction satisfaction before deciding whether to refine, expand, or permanently integrate the feature into the mainstream Discover experience.

While the current phase of the experiment is limited to video content, the structural logic of the Dive deeper framework makes it easily adaptable to other media types, including standard text articles, breaking news alerts, and long-form investigative reports. Google has not yet provided an official timeline for scaling the test beyond videos or introducing it to broader international markets and non-English languages.

As the test progresses, webmasters, SEO analysts, and publishers will be watching closely for updates regarding Search Console reporting transparency and official guidelines on how to optimize for AI-driven discovery feeds. For now, the Dive deeper experiment stands as a clear signal of Google’s ongoing commitment to transforming static content feeds into dynamic, AI-moderated exploration hubs.