Ad-hoc network such as wireless sensor networks (WSNs) and the Internet of Things (IoT), require key network-related information to be broadcast for efficient functioning. The communicating nodes in these networks are energy constrained and must operate in a distributed manner. Simple broadcast mechanisms such as flooding, are not energy efficient. Probabilistic forwarding has been employed as an alternative, wherein a node forwards a previously unreceived packet with some probability p. While this performs better than flooding in terms of the number of transmissions, reception of packets is not guaranteed. We propose encoding information at the source along with probabilistic forwarding. We show that on most well-connected graphs such as grids, lattices and random geometric graphs, we can have a large fraction of nodes to receive the information from the source while having lesser number of transmissions compared to probabilistic forwarding with no coding.