Readers can now shape what appears in Google Discover by typing plain-language instructions directly into the feed, giving them more control over the topics, stories, and links that show up. The change turns the recommendation feed into something a user can steer in real time, rather than a stream shaped only by past clicks. The update matters for publishers because the same signals that influence what readers see also shape what audiences can discover from any given site.<\/p>\n
Inside Google Discover, a three-dot menu on individual cards opens a way to describe, in your own words, the topics or sources you want more of, and the ones you want less of. Google processes that natural-language input and adjusts the feed on the fly, while remembering your requests for future sessions. The result is a feed that aims to match what you care about right now rather than relying solely on inferred behavior from prior reading.<\/p>\n
The instructions are written conversationally. A reader can ask to see more coverage about a niche hobby, request fewer results from a particular publisher, or signal interest in a specific type of story. The wording is up to the user; the system reads the request and applies it to the recommendation stream without requiring a fixed set of categories or filters.<\/p>\n
Once a request is logged, Discover updates the recommendations it surfaces during that session and carries the preference forward. This means a single written instruction can keep shaping results across multiple visits, not just the next scroll. For users, the practical gain is faster alignment between what they ask for and what shows up.<\/p>\n
Looking at the kinds of refinements readers make can show angles editorial teams might miss in areas where search volume is low but reader interest is evident.<\/p>\n
When users tell Discover they want more of a very specific subject, those signals can help a content team identify emerging interests before they appear in standard keyword tools. A reader asking for more deep dives on a narrow industry topic, for example, is a strong hint that an underserved audience exists for that angle.<\/p>\n
Feeds that match reader preferences can make it more likely that a publisher’s content reaches readers who have asked to see it. A story aligned with a reader’s stated preferences may be more likely to be surfaced than one relying on broad behavioral signals alone, which can help them move from a topic idea to a published story read by a relevant audience.<\/p>\n
The shift moves Discover closer to a feedback loop than a one-way broadcast. Instead of training the algorithm only by clicking, liking, or ignoring stories, readers can now describe their intent directly, in language they choose, and watch the feed shift in response. The core promise is a feed that listens to stated preferences as well as observed behavior.<\/p>\n
Conversational search in Google Discover is a feature that lets readers type plain-language instructions through a three-dot menu on feed cards, asking to see more or less of certain topics or links. The feed adjusts to those requests in real time and remembers them for future sessions.<\/p>\n
Open the three-dot menu on any card in Discover and write what you want to see more of, or less of, in your own words. Google applies the request to your current feed and carries the preference forward.<\/p>\n
Content teams get a new signal for niche reader interests because users can ask for specific topics, and a publisher’s content is more likely to reach readers who asked to see it.<\/p>\n