The failed version did its job if it taught the successful one something essential.

A version that failed can still be the reason the next one works. Prototypes are supposed to fail; that is their function, and judging them as though they were meant to succeed misreads what they were for. Whatever the failed attempt taught you is what it was actually producing.

When something you made does not work, ask what it revealed that you could not have known without building it. Count that as its output. Some attempts exist to inform the ones that follow.