Accurate decoding of external signals in presence of cellular noise is crucial to ensure appropriate cellular responses. Although multiple feedback loops are likely to influence the information flow in signalling cascades, their effects on information decoding as well as the associated cost-benefit trade-offs remain largely unexplored. Here, we investigate the role of the feedback regulation within the MAPK phosphorylation cascade that mediates the pheromone response in Saccharomyces cerevisiae. We observe that several negative feedbacks that are induced at different levels of stimulation progressively reduce the background activity and suppress noise within the pathway, thus improving accuracy of signal decoding. However, this improved accuracy comes at an energetic cost of higher ATP consumption. In contrast, the positive feedback on the MAP kinase Fus3 reduces energy consumption. Our computational analysis demonstrates that the strengths of these feedback loops are tuned to optimize the tradeoff between information transmission and energetic cost. Taken together, our study demonstrates how design of signaling cascades can be understood based on optimal signaling accuracy.