In Industry 4.0 scenarios, novel applications are enabled by the capability to gather large amount of data from pervasive sensors and to process them in order to devise the “digital twin” of a physical equipment. The heterogeneity of hardware sensors, communication protocols and data formats constitutes one of the main challenge toward the large-scale adoption of the Internet of Things (IoT) paradigm on industrial environments. To this purpose, the W3C Web of Things (WoT) group is working on the definition of some reference standards intended to describe in a uniform way the software interfaces of IoT devices and services, and hence to achieve the full interoperability among different IoT components regardless of their implementation. At the same time, due also to the recent appearance of the WoT W3C draft, few testbed and real-world deployments of the W3C WoT architecture has been proposed so far in the literature. In this paper, we attempt to fill such gap by describing the realization of a WoT monitoring application of a generic indoor production site: the system is able to orchestrate the sensing operations from three heterogeneous Wireless Sensor Networks (WSNs). We describe how the components of the W3C WoT architecture have been instantiated in our scenario. Moreover, we demonstrate the possibility to decouple the mash-up policies from the network functionalities, and we evaluate the overhead introduced by the WoT approach.