Google Discover is testing a new “Dive Deeper” button that turns a single video card into a broader topic doorway. Google’s Robby Stein, VP of Product for Search, confirmed the experiment on X on Friday, saying the goal is to make topic exploration and discovery of related web content easier in Discover.
What the test looks like
The test starts with videos. In Google’s example, a Discover card shows a KCRA 3 weather video about a Northern California heat wave. Under the video, a short summary carries the label “Generated with AI, which can make mistakes.” A “Dive Deeper” button sits below that summary.
Tapping it opens a second screen with the same video, a longer summary, and a row of source cards. The first card links to kcra.com. A third screen, “Explore more stories,” groups short summaries under sub-topic headlines, each followed by source links and descriptions of the linked coverage.
Stein described the experience as a “short topic overview with prominent links to related stories, community reactions & original reporting.” Google will test multiple designs in the weeks ahead, but it has not disclosed the countries, languages, or test size.
Why this matters for traffic teams
Discover already sends meaningful non-search traffic to publishers and e-commerce content teams. This test adds another interactive layer: one video card can become an entry point to multiple sources. If your content is not part of the related coverage, you may be left out of the click path.
Google rolled out AI previews on trending Discover topics in October 2025 in the U.S., South Korea, and India. A generative AI performance report for Discover became available to all websites worldwide on Aug. 31, according to Google. The report counts impressions of links shown in generative AI features in Discover, grouped by page, country, and date. It does not break down performance by feature, and Search Labs experiments are excluded.
That means impressions from a “Dive Deeper” overview, if counted, would be mixed in with other AI features. Attribution will be fuzzy, so don’t wait for perfect reporting before acting.
How to prepare now
Treat each video as the start of a topic journey, not a standalone asset. Here’s where to focus:
- Build topic clusters around hero videos: original reporting, supporting articles, and community roundup pages.
- Make video titles and thumbnails clear enough to earn the initial Discover card.
- Strengthen internal links so Google can surface your related coverage inside topic overviews.
- Monitor the Search Console generative AI report for Discover, but treat it as directional rather than feature-specific.
A practical framework is to map one “hero” video to at least three supporting pieces: a source article, a reaction or community post, and an original report. That mirrors the links Google is showing in this test.
The bigger Discover picture
Google is experimenting with designs, not committing to a final layout. The test may stay limited to videos for now, but the direction is clear: Discover cards are becoming topic hubs rather than isolated content units. For marketers and agency teams, the win goes to those who combine strong video with deep topical authority.
Source: Search Engine Journal
