Soft pneumatic actuators are very popular in the soft robotic community due to their ease of manufacturing and simplicity of control. Currently, the efficiency of such soft actuators and their ability to do useful work are rarely investigated in a formal approach. The lack of task-orientated development approaches presents a barrier to utilize soft robotic systems in our everyday lives. In this paper, we describe an experimental approach based on port-Hamiltonian theory applied on a type of pneumatic network (pneu-net) actuator to investigate the efficiency of task-orientated work. We can obtain efficiency from the external interactions of the port-Hamiltonian system. If we can minimize the internal energy interactions, then the power continuous nature of the port-Hamiltonian structure ensures more input energy will result in more useful work done at the output. We found out that higher efficiency actuators can be achieved with a softer material and a thinner wall thickness in the desired direction of the deformation. The internal mechanical energy storage is reduced as a result. However, if the task requires a higher work-done then a stiffer material is required. We can start to define a design approach based on the task. The task can be generalized in terms of energy. We can select the material properties suitable for the magnitude of work done. We can design the geometry to minimize the internal energy stored. The empirical model of the port-Hamiltonian structure provides insights into how the mechanical efficiency varies in terms of design parameters and the port-Hamiltonian approach is a step towards more practical, task-orientated soft robotic systems.