Design Iteration: 100% Necessary
It might be common wisdom that you need to iterate your design.
During the development of the Draftsman multi-tool, as soon as a model had been finished in CAD it was 3D printed. This gives you something tangible within the hour that you can play around with and get the "feel" of it.
In a perfect world, you would design your widget and account for everything and make it, and it would be glorious. However, the problem is that people might end up having to use your widget, and touch it with their hands. Even though 3D prints don't have the same feel as a metal part, they are still very handy when people's hands, tools, machines, anything else will be involved that is not in your CAD model. An original design point of the Draftsman was to incorporate a tiny calipers for measuring accurately on the fly. This looked great in CAD and on paper, and the prototype functioned great with it, however it was quickly realized that you would pinch your finger in the calipers when you closed it, 100% of the time.
The design was adjusted for the second concept, and now a depth gauge was added, preventing your finger from slipping into the notch where it would get caught, and adding more functionality. Except now you would stab yourself with the gauge whenever you opened the calipers, 100% of the time. The funny thing was I had even thought of this while modeling it, but decided that people (including myself) would just hold it differently to use the calipers (They can't by the way, and I couldn't either). So even the solution to the problem became a problem, because CAD is an intangible process. I eventually decided to drop the calipers, because I felt the value added did not outweigh the complexity it added.
Since these prototypes were 3D printed and readily available to play around with, these iterations were able to happen within a few hours instead of a few days or weeks waiting for metal parts. Not only that, it allowed very tangible feedback for me and maybe even more importantly allowed other people to test it too. Feedback is good feedback, and you can use it to design a better product.