Recommendation systems have traditionally relied on short-term engagement signals, such as clicks and likes, to personalize content. However, these signals are often noisy, sparse, and insufficient for capturing long-term user satisfaction and retention. We introduce Retentive Relevance, a novel content-level survey-based feedback measure that directly assesses users' intent to return to the platform for similar content. Unlike other survey measures that focus on immediate satisfaction, Retentiv...