While two-sided platforms (e.g., video game consoles) depend on complements (e.g., games) for their success, the success of complements is also influenced by platformlevel dynamics. Research suggests that greater platform adoption benefits complements by providing more potential users, but this assumes that platform adopters are homogeneous. We build on extensive research exploring the heterogeneity between early and late platform adopters to identify counterintuitive dynamics for complements. Complements launched early in a platform's life cycle face an audience entirely of early platform adopters, whereas later-launching complements face a mixed audience of both early and late adopters, and we argue that differences in preferences and behavior between early and late adopters affect whether complements will succeed and which types will be most successful. We explore these dynamics in the context of the console video game industry using a unique data set of 2,918 video games released in the United Kingdom from 2000 to 2007. We show that despite the increase in the potential user pool as the platform evolves, video games launched later in the platformlife cycle realize lower sales than those launched earlier. While increased competition explains part of this effect, we show substantial evidence consistent with our theory of preference differences between early and late adopters. This includes the finding that the negative effect is stronger for novel games and that the gap between popular and less popular complements widens as later adopters move into the platform, consistent with late adopters being risk averse and seeking to avoid purchasing mistakes.

Additional Metadata
Keywords Complementors, Demand-side view, Technology evolution, Two-sided platforms, Video games
Persistent URL dx.doi.org/10.1287/orsc.2017.1183, hdl.handle.net/1765/106071
Journal Organization Science
Rietveld, G.J, & Eggers, J.P. (2018). Demand heterogeneity in platform markets: Implications for complementors. Organization Science, 29(2), 304–322. doi:10.1287/orsc.2017.1183