@dch Chesterton's Fence applies for sure, but my little rule focuses more on the social aspect.
Basically, I noticed with process automation, the more your flow differs from how folks already do things, the less folks will want to use it. You need to spend more time and money simply forcing people to adopt your idea.
Lots of startups fail and big companies embarrass themselves by ignoring this (Windows 8 is an easy example).
@dch my automation approach is to learn the existing process—by doing if possible—talk to folks getting work done, document process, help sand off rough edges, and then *maybe* write code to automate things. Always checking in to see if my changes are helping.
Turned a punch card system into a shared spreadsheet into a Rails app over six months with this approach. (A Rails app they kept using for 10 years after management said they'd stop using it, because mgmt ignored what I was just saying)
Random Geek