Current quantitative evaluations in various research areas for publish/ subscribe systems use artificially created event messages to model the system workload. The assumptions made to create these workloads are rather strong and hardly ever described in detail. This does not allow for a repetition of experiments or comparative evaluations of different approaches by different researches. In this paper, we present an evaluation of the distributions of the values of attributes typically used in online auction scenarios. In particular, we focus on auctions of fiction books. We further show our approach of creating event messages by the help of the gained information. Publishing this information on how to create a typical workload for online auctions should allow for the repetition of experiments and the comparison of different evaluations.