Market experimentation of disruptive innovations under network effects

Because disruptive innovations rely on untested business models, a real market experimentation plays a key role in evaluating their probability of success. However, the design of a real market experimentation is particularly difficult to establish when the business model underpinning the innovation displays strong network effects. On the one hand, minimizing direct costs and sunk costs in case of failure, induces the innovator to keep the experimentation as small as possible. On the other hand, conducting the market test on a too low scale entails the risk that network effects do not fire up, because a critical installed base of users is not reached. This trade-off is an important issue, as network effects are pervasive in today's economy. In this article, we develop an economic model that examines whether the adoption rate on the whole market can be extrapolated from the adoption rate on the market test, depending on the size of the market experiment. We show that below a minimum threshold, the innovator may not properly assess the real market potential of an innovation, and may miss the opportunity of progressively refining the business model underpinning the innovation. This threshold is inversely related to (i) the intensity of network effects and (ii) the fraction of consumers who are sensitive to the network effects of the new product or service. Conversely, we demonstrate that beyond a certain threshold there is no benefit in extending the size of the experiment. We derive from the model managerial implications regarding the management of a market experiment under network effects.