GET RICH OR DIE TRYING… UNPACKING REVENUE MODEL CHOICE USING MACHINE LEARNING AND MULTIPLE CASES RON TIDHAR Department of Management Science and Engineering Stanford University
[email protected] KATHLEEN M. EISENHARDT Department of Management Science and Engineering Stanford University
[email protected] December 7, 2018 Revise and resubmit, Strategic Management Journal Special Issue on Question-Driven Research ABSTRACT While revenue models are gaining strategic importance, related research is incomplete. Thus, we ask the phenomenon-driven question: “When should particular revenue models be used?” We use a novel theory-building method that blends exploratory data analysis, machine learning, and multi- case theory building. Our sample is from the AppStore, an economically important setting in the digital economy. Our primary contribution is a framework of configurations of effective revenue models. It indicates new theoretical relationships linking quality signals, user resources, and product complexity to the choice of revenue model. It also unlocks equifinal paths and new revenue models (e.g., bundled and fragmented). Overall, we contribute a more accurate and theoretical view of effective revenue models. We also highlight the surprising complementarity of machine learning and multi-case theory building. Keywords: Revenue models, competition, mobile application products (apps), machine learning, multi-case theory building. INTRODUCTION In 2011, music streaming service, Spotify, launched its product to U.S. listeners. Although similar to other music streaming services, Spotify was notably different in its revenue model. While other companies relied on either a paid model (e.g., Apple iTunes) or an advertising one (e.g., Pandora), Spotify offered its service for free, but included a paid premium version.