In this article we study the use of intention surveys to predict the effects of a possible entrant. The case under investigation deals with the introduction of private broadcasting in the Netherlands. Several predictions of the advertising expenditures in various media are given which depend on a number of scenarios. These scenarios are used to reduce the discrepancies between behavioural intentions and actual behaviour. The predictions of the most realistic scenario are compared with their realizations, and the differences are analyzed. To this end the prediction error is decomposed into an intention error and a sampling error. This decomposition offers good opportunities to analyze discrepancies between intentions and actual behaviour.