Nice article, and I agree with the core idea of it, but I would argue, that the approach you took with taking weeks or months to setup databases and design DW etc. is not optimal and indeed this approach requires you to have great soft skills, since you need to convince managers that at the end everything will be ok :)
I would argue for more iterative approach — where you make minimal contribution to every part of the stack, but already deliver value to the team with models (maybe only baseline models) running in production within weeks.