It depends upon your intended output and subject. Animation for Games, and video to a lesser extent require highly optimized topology, and may need to be designed in a specific way to accommodate some specific production workflow. Working for print isn’t overly concerned with topology, since most stuff is just going to be decimated anyway. In terms of subject, organic detail is more easily sculpted than modeled, and hard surface stuff is always going to be easier and more precise to do with polygonal modelling.
It used to be that modeling a base mesh in a traditional polygonal modeler and importing into a sculpting tool for fine detail was the way most people worked. As sculpting and retoplogy tools have improved, it’s now also common for people to retopologize models that have been sculpted from scratch.
Basically, just figure out what you want to do specifically, start doing it, and you’ll discover what works best for your workflow and intended output.