The three-question test we run before automating anything

Automation is a tool, not a goal. Before we build anything, every candidate process gets run through the same three questions.

First: does it happen often enough to matter? A task done once a quarter usually isn't worth automating, no matter how annoying it is in the moment.

Second: is the process actually consistent? Automating a workflow that changes every time just moves the chaos somewhere less visible — it needs a repeatable shape first.

Third: what happens when it breaks? If a silent failure would cost more than the manual version ever did, the automation needs monitoring built in from day one, not bolted on after something goes wrong.