Pretty much this. It's especially true in startups where new features and edge cases pile up really fast. Solve a new edge case manually in the beginning (like

Pretty much this. It's especially true in startups where new features and edge cases pile up really fast. Solve a new edge case manually in the beginning (like adding a simple field instead of a smart calculation, like you mentioned) and work from there. It might turn out, in a few months, that that new feature nobody uses anyway so it's dropped. This is easier if you obey: 6) Think at product level