Success Prediction and Release Strategy Optimization for Independent Musicians

Abstract

In an era where digital platforms have democratized music distribution, small and emerging artists face overwhelming competition and a lack of data-driven frameworks for strategic decision-making. Thus, we address two core questions: (1) Which features of artist best predict future success? and (2) How can emerging musicians optimize their release strategies to maximize audience growth?

To answer the predictive question, we train interpretable machine learning models—including decision trees and logistic regression—on Spotify metadata for over 9,000 emerging artists. We find that recent release frequency and a high proportion of singles in an artist’s catalog are strong predictors of follower growth, with the best models achieving F1 scores over 0.80.

With the knowledge that the frequency of single releases are predictive of artist success, we then formulate a prescriptive framework to optimize artist release strategies. We model the daily follower growth of an artist over time as the sum of exponentially decaying functions triggered by single and album releases. We define these functions in terms of the time between successive releases, an artist's follower count, and the number of new tracks featured in an album. Using non-linear least squares models, we fit these functions using follower time-series data from Songstats. Incorporating these functions into a mixed-integer nonlinear program, we then solve for optimal single and album release schedules over a fixed planning horizon.

Our model's optimum solutions recommend the release of as many singles as possible, spaced evenly across a desired release period to maximize follower growth. Conversely, our model places less importance on saving unreleased tracks for an album release. This result highlights a functional distinction between the two mediums: singles are particularly effective at driving audience expansion for smaller artists, while albums are better at generating revenue for established artists.