From zero to hero - how to balance lost returns and wasted investments using predictions of rare events

This paper addresses decision problems associated with rare events, which arise in the context of resource allocation within a set of products, for example, music artists who recently signed up on a user-generated content network. Among these artists only few will eventually become successful in the long-term as success is a rare event and thus per se not likely to occur. Hence, for record labels investments in all artists are associated with an extreme waste of resources, whereas selecting too few artists comes at the cost of lost returns. The above-described trade-off concerns not only record labels but also multinational companies with broad and deep product pipelines or venture capital funds investing in start-ups. We propose a novel managerial framework how to maximize profits when deciding on how many as well as which recently launched products, start-ups, or upcoming music artists to select in an early stage of the life cycle, namely by solving the trade-off between wasted investments and lost returns. The profit maximization draws on the available information, the respective return on investment as well as the chosen prediction model, which makes it a suitable tool for managers. We apply our framework in the context of SoundCloud, the world’s leading user-generated content network in the music domain, and make recommendations in which artists to invest shortly after sign-up