Good design, experience, analysis, and simulation is a fantastic tool for reducing design cycle iterations required to go from prototype to product, but it is not enough without testing.
The testing phase is critical because it can either make or break a design. If the design breaks, sometimes it is an easy production fix. Sometimes if the design breaks, it is unfixable and a new architecture is needed. You might think this is bad, but it isn't. Nothing could be worse than mass-producing a product that will consistenly fail when deployed or in use.
I always fully test products to verify assumptions and all uncontrollable externalities are factored into the testing before approving for production.