Use either power or energy to calculate efficiencies. Your choice is determined by a focus on either the power in pathways for processes that are still happening, or on the energy that has accumulated in the stores during the process.
Efficiencies are useful as guides to action because they show how far short one has fallen of perfection (100 % efficiency in almost all cases). The difference between what you have achieved and what could be achieved may help you to see whether it's possible to make significant improvements. You might seek to develop what you are working on, or to adopt a completely new approach. In either case you'll be comparing your calculated efficiency with other approaches, or perhaps even with a theoretical maximum, calculated from first principles. (We don't show you how to try to do that here, but the principles of thermodynamics are often useful. These take the story of energy developed here even further.)
Energy efficiencies, using calculations based on stores, will be more general than power efficiencies, based on pathways. The latter incorporate more information about the devices. That's why they're more specific and often better suited to development work by engineers. However, the most general approach, using energy descriptions, often has no connection with the mechanism at all (a distinction made much of in the Energy sequence). This makes comparisons possible: energy is a universal non-arbitrary currency, which allows the outcomes of completely different physical options to be compared reliably and objectively.