The management of data is crucial in today’s organizations, making it necessary to specify exactly how data is created, accessed, and manipulated during business process enactment. Given the importance of data, it comes as a surprise that approaches like BPMN only provide limited support for modeling data and how it is read and written. In particular, they cannot represent multiple data objects of the same type, and they lack concise semantics for multi-instance data objects. Against this background, this paper proposes an extension to BPMN process models by introducing variable identifiers to distinguish individual data objects of the same class in a given process. The behavior is detailed using translational semantics to Colored Petri nets, and a set of verification mechanisms is presented that allow for a more precise analysis of data objects in business processes.