Companies are under tremendous pressure to analyze and optimize their productional and logistical networks in today's global business world. Hence, practitioners and researchers show great interest in digital twins. A digital twin is a virtual construct that mirrors real-world objects and conceptual ideas while it processes, handles, distributes, and optimizes data streams. Its main purpose is to visualize, analyze, and optimize objects and systems, making a digital twin highly suitable to help companies gain an advantage over their competitors through a great degree of transparency over their production and logistics. Therefore, almost every company evaluates the usage of digital twins. Nevertheless, many companies struggle to instantiate digital twins since they lack fundamental knowledge about all necessary components of a digital twin and the individual requirements for the operation of the digital twin. This lack of knowledge hinders the broad implementation in practice. Research shows many descriptions of theoretical use cases and field studies but rarely describes digital twins in real operational settings. To address this research gap between theoretical concepts and practical challenges of the implementation of digital twins, this paper investigates the practical requirements of digital twins in real-life usage. Based on a thorough interview series with international manufacturing and logistics experts, we identify and analyze the requirements for data handling, data policy, and services of digital twins and cluster them according to the requirements engineering approach. Through a comprehensive overview of the different requirements, the paper delivers profound insights into the needs of companies from various fields and, therefore, gives practitioners a guideline on crucial aspects of implementing digital twins.