TechPrufer
Get started
Spotify
Spotify logoSpotify · Feature trail

Taste Profile Personalization Control

2 posts · Dec 2022 – Sep 2023

What this trail covers

Spotify has introduced a feature allowing users to exclude specific playlists from influencing their taste profile and recommendations. This addresses the issue of 'functional listening' (e.g., for sleep, parties, or children) and shared account listening, which previously skewed personalization. The system integrates this exclusion logic into core taste profile systems, impacting various recommendation surfaces like Home, Discover Weekly, and Radio. The feature was developed iteratively, starting with a focus on user control over personalization.

All Spotify trails
  1. Exclude from Your Taste ProfileThis post details the technical implementation of the 'Exclude from your taste profile' feature. It explains how the feature impacts core taste profile systems (user embeddings, recent listens, top tracks/artists, interactions) to prevent functional listening from skewing recommendations. The implementation strategy involved applying filtering logic directly within these systems to minimize changes required by downstream recommendation surfaces. It also describes the A/B testing methodology, which focused on editorial playlists and measured the impact on recommended content consumption and overall user engagement. The post highlights the decision to exclude plays from excluded playlists (both past and future) and to not infer other content types for exclusion, prioritizing explicit user control and implementation simplicity.
  2. What’s a “Listening Personality”?This post introduces the concept of 'Listening Personality' as a new feature for Spotify Wrapped. It defines four binary attributes (Familiarity vs. Exploration, Timelessness vs. Newness, Loyalty vs. Variety, Commonality vs. Uniqueness) that describe how users listen to music, independent of their genre preferences. The post details the technical calculation of these attributes, including the use of 'affinity' scores for artists and popularity metrics for songs. It explains how these attributes are normalized across the user base to generate a four-letter code representing a Listening Personality. The post also touches on the potential for users to reflect on and change their listening habits based on their personality.