There are rigorous standards of when does a QC work correctly, not too different from how classical computers are validated. On top of that, there are many metrics to determine how much faster QCs are at given tasks, and how improving that is a vibrant area of research.