I think this approach is more common than the hype for actual work. I do something similar, many back and forth, then settle on something often with now known tradeoffs, written by hand to spot issues as a final guard/ keep consistent naming etc.