The effect of network structure in industrial diffusion processes

Abstract

The development and diffusion of innovations involves groups of organizations with many different roles, who interact with each other (suppliers, adopters, third parties, etc.). This paper examines the nature of the communication networks that exist between adopting organizations, and from third parties and suppliers to adopters. Quantitative models of diffusion often assume that information about innovations flows along pre-existing links, that this information flows from adopters to non-adopters directly, and that there is perfect mixing in the population (i.e. every actor has an equal chance of communicating with every other actor). These are strong assumptions which should be subject to testing. We investigate the impact of differing network topologies of communications and alternative models of social contagion on observed adoption patterns. Network topologies are also extended beyond the focal industry to include suppliers, consultants and other customers. We use a simulation model to test the effect of departures from traditional assumptions on the diffusion pattern and undertake an empirical field study to examine the prevalence of those departures in one specific industry. Our simulation findings suggest that network structure can have a substantial effect on the manner in which innovations diffuse, while innovation-specific communication links, and communication through third parties does not change the shape of the penetration trajectory as much as it alters the places in the process where salesforce effort can offer the most leverage. The empirical study shows strong evidence of imperfect mixing, that both pre-existing and innovation-specific communication links are used, and that communication through third parties may be as important to the diffusion process as direct links from adopters to potential adopters.