Most automation projects fail for the same reason: they start with the most visible process instead of the most suitable one. The process everyone complains about is often too messy, too rare, or too full of judgement calls to automate well.
A better starting point is a simple score. For each candidate process, rate four things from one to five: how often it happens, how rule-based it is, how stable the systems involved are, and how costly mistakes currently are. Multiply them together. The highest scores are your first wins.
Volume matters because automation savings compound. A ten-minute task done fifty times a day is worth far more than a two-hour task done once a month. Rules matter because bots are excellent at following them and poor at judgement. If people regularly argue about the right answer, fix the process before you automate it.
System stability matters because bots that click through screens break when those screens change. If an application is about to be replaced, automate through its API or wait. And error cost matters because automation's quiet superpower is consistency: a bot that keys invoices never transposes two digits.
Start with one process, measure the before and after honestly (hours, errors, turnaround time), and use those numbers to choose the next. A small automation that clearly works beats an ambitious one that half works, every time.
