When do you think an USA company should transition from off-the-shelf software solutions to custom development? How important is it to first identify the most labor-intensive processes and only then decide which ones are truly worth automating, to avoid creating an unnecessary system from multiple disparate services?