YouTube adds a “custom feed” to its home page that lets users prompt it to create a constantly refreshing feed, available to signed-in US users with history on
YouTube isn't exactly replacing the traditional search bar, but it is supplementing it with a way to watch a “custom feed” of videos based on your nuanced prompts.
Context & Ripple Effects
YouTube has repeatedly added ways to shape recommendations, from topic controls and channel-level “For You” sections to homepage redesigns and options to suppress channels. The new home-page control extends that arc from selecting among preset signals to describing the feed a viewer wants in their own words.
It also follows Google’s move to test conversational video discovery through Ask YouTube. Together, the changes make natural-language input a more prominent layer on top of YouTube’s established recommendation system.
First-order effects
- Signed-in US viewers with watch history can create a continuously refreshed home feed using nuanced prompts, rather than relying solely on the default homepage recommendations or conventional search.
- YouTube gains another first-party preference signal: the wording of a prompt can indicate viewing intent more explicitly than passive watch history alone.
Second-order effects
- Creators’ discovery may become more dependent on how well their videos match prompt-defined interests, alongside the signals that drive placement in the standard recommendation feed.
- A prompt-driven feed can give viewers a more direct way to steer away from undesired generic recommendations; its usefulness will depend on whether the resulting feed consistently reflects those instructions.
Third-order effects
- If YouTube expands this approach, video discovery could shift from a largely implicit, algorithm-led homepage toward a hybrid model in which users actively specify intent while recommendations handle ongoing replenishment.
- The move reinforces competitive pressure on video platforms to combine search, recommendations, and conversational interfaces rather than treating them as separate entry points.
The trend: This is one data point in the shift from passive personalized feeds toward conversational, user-directed recommendation systems.