Internal tools are a margin decision

A scheduling tool that saves a supervisor fourteen hours a week is not a software expense. It is fourteen hours of capacity you already pay for and currently waste.

The evaluation is simple: hours recovered times loaded hourly cost, against build plus three years of maintenance. If the payback is under twelve months, the interesting question is why you have not built it yet.

The trap is building the tool the loudest person described instead of the one the process needs. That is why we spend the first week watching the work rather than gathering requirements in a meeting room.