Wireless sensor networks (WSNs) have long been deployed for single application only; a single gateway collects data of singular application from multiple source nodes. However, recently the concurrent usage of WSNs for multiple applications has been start frequently. At the topological level, a consequent migration from single gateway to multiple gateways has taken place, wherein each gateway belongs to a unique application. Also, as known clustering sensor nodes to aggregate, and filter the data at cluster heads prior to sending optimizes the communications cost. However, clustering WSN-supporting multiple applications results into heterogeneous cluster. Where, the current literature does not appreciate this migration to be a deviation from existing communication paradigms. It assumes that the state-of-the-art in unicast and multicast routing render acceptable performance. In this paper, we argue that maintaining such status quo is weak, as it results into inefficient usage of current path discovery, routing and data delivery protocols, resulting into (a) broadcast storms and collisions due to cascaded route discovery messaging in the form of RREQs for multiple destinations as in unicast routing protocols and (b) an inability to route aggregated-multiple-destination-oriented data of different types. In this paper, we first identify the motivation and need for a new communication paradigm, which differs from both unicast and multicast scheme, and then present Carrefour casting as a new communication paradigm. We propose Carrefour Routing for Cluster-based WSN (C???C) to discover, route, and deliver data to multiple gateways concurrently. C???C includes two mechanisms, (a) Aggregated route discovery mechanism to optimize and reduce the path discovery overhead, and (b) Carrefour Data Combination (CDC) to optimize the data gathering and dissemination. C???C is easily implemented through minimalist modifications to the data structures of AODV and data packets, and uses - routing engine of AODV. The evaluation of our implementation shows better results than AODV in route discovery overhead, end-to-end delay, and communication cost. Overall, the simulation results demonstrate the effectiveness of C???C as the number of applications is increased.