Research
A UX research study examining how listeners discover new music on Spotify — and where the platform's recommendation systems support or undermine genuine exploration.
Spotify's recommendation engine is one of the most sophisticated in consumer software. But recommendations optimized for engagement don't always feel like discovery. This study asked: what do listeners actually experience when they try to find new music they love — and what gets in the way?
Over six weeks, I recruited 10 participants across different listening profiles (casual, enthusiast, niche-genre) and combined diary study entries with semi-structured interviews to understand their discovery journeys in depth.
Participants logged their listening experiences three times a week for four weeks, noting when they discovered something new, how, and how they felt about it. Sessions were followed by hour-long interviews using a protocol designed to surface mental models around "discovery" vs. "familiarity."
Screen recordings captured actual in-app behavior, which often diverged from what participants reported in diaries — a productive gap that became one of the study's central tensions.
"I always say I want to find new stuff, but when I open the app I just go back to what I know."
Discovery and listening are different modes. Participants rarely discovered and listened in the same session — discovery required a particular kind of open, exploratory attention that didn't coexist with background listening.
Algorithmic recommendations were trusted but not loved. Participants relied on Discover Weekly but described it as "safe" — rarely surprising, rarely emotionally resonant in the way a friend's recommendation could be.
Social discovery was underutilized but highly valued. When participants did discover something via a friend, collaborator, or live show, the emotional attachment was significantly higher than algorithmically-surfaced music.
Genre boundaries created friction. Several participants in niche genres found that the recommendation engine would surface popular-adjacent music rather than deep cuts — flattening their listening identity over time.
Based on findings, I developed three design recommendations for Spotify's discovery experience: (1) a distinct "exploration mode" that signals to the user — and the algorithm — that they are open to unfamiliar territory; (2) strengthening social sharing flows so friend recommendations are surfaced more prominently in discovery contexts; and (3) giving users more visibility and control over the signals that feed their taste profile.
These recommendations were presented as a research report with annotated wireframe sketches illustrating possible UI directions for each, though the project scope was research rather than design execution.