By running the code. Let's say the code is bad. The software will let you know if there's a logical error. The software will be hard to maintain when you want to update it, The software will be hard to use when users try it. There's no feedback like the one the software gives. The proof is in the pudding. If you're new to software and a junior developer, then outside feedback is much needed, but once you're experienced. The most useful feedback is going to be the one the software gives you.