The most expensive sentence in software is 'we will just build it'. Sometimes building is right. Often it is not.
Signals that building makes sense
- the process is important and repeated many times
- existing tools genuinely cannot be configured to fit
- the manual work costs more than the tool would
- there is a clear, small first version
Signals to stop
- the requirement is still changing weekly
- an off-the-shelf tool would do 90% of the job
- nobody owns the ongoing maintenance
- the first version has grown to include everything
We aim for a small MVP with a clear definition of done. If a process is better served by configuration, we will say so. Building is the last option, not the first.