Product Improvement for YouTube's What-to-Watch Suggestions

YouTube suggests videos to people throughout the app. How would you make those suggestions better? ```hint Decide what "better" means Better could mean more minutes watched, happier viewers, more variety, or a fairer chance for new creators, and these pull against each other. Choose one group of viewers and one definition of better before proposing changes. ``` ### Clarifying Questions - Which surface is in scope: the home feed, the suggestions beside a playing video, Shorts, or all of them? - What does the team mean by better: more watch time, higher viewer satisfaction, discovery of new topics and creators, or fewer harmful suggestions? - Which viewers matter most: signed-in regulars, new viewers with no history, or households sharing one account on a TV? - Can the answer include controls viewers use to steer suggestions, or only changes to the ranking itself? ### What a Strong Answer Covers - A definition of better recommendations that weighs viewer satisfaction against raw engagement, tied to YouTube's business and to creators. - One chosen viewer segment and the specific ways recommendations fail it today. - Concrete improvements, covering both ranking signals and viewer-facing controls, prioritized by impact and effort. - A primary metric that reflects satisfaction rather than clicks alone, with guardrails for watch time, creators and content safety. - Risks such as clickbait incentives and rabbit holes, and how the changes would be tested. ### Follow-up Questions - Your change improves satisfaction scores but lowers total watch time. Do you ship it, and how do you decide? - How would you tell whether added exploration helps viewers find new interests or just shows them videos they do not want? - Small creators say they never get recommended. How would your design respond without lowering quality for viewers?