This paper describes an experimental approach towards the definition of a metric for integration test effort based on enterprise resource planning (ERP) system logs. By nature, ERP systems are complex. After implementing an ERP-system, it is typically subjected to many, and often concurrent, changes. On the one hand, these factors that cause the effort, which is to be invested in integration testing, become exceptionally hard to predict. On the other hand, the effect of invested integration test effort can seldom be determined for the same reasons. To show how aspects of effort can be distilled from the flow of changes, we propose a change stream reconstruction process based on sap r/3 (SAP AG, Walldorf, Germany) system logs. Next, we introduce sampling methods yielding multivariate datasets, which bring the following results. The number of carriers of the change and their testing lead-time correlate moderately. The number of objects comprising the system's code and configuration, which have been touched during the change process, and lapsed testing lead-time are not correlated. Six independent cases are subjected to two distinct lines of investigation.