Not every manual process is worth automating, and the processes most worth automating aren't always the most obvious ones. A useful starting point is identifying where manual work is both frequent and error-prone, rather than automating whatever seems most technically interesting.
How to identify a good first candidate
- It happens often enough that the time saved adds up — daily or weekly, not once a quarter
- It follows a consistent, rule-based pattern rather than requiring judgment each time
- Manual errors in this process have caused real problems (missed follow-ups, data entry mistakes, delays)
- It currently depends on one person remembering to do it
Common early automation targets
Invoice generation and follow-up, data entry between systems that don't currently talk to each other, routine notifications (low stock, overdue tasks, new form submissions), and recurring reporting are common starting points because they're rule-based, frequent, and directly tied to time savings that are easy to measure.
What automation won't fix
Automation makes an existing process faster and more consistent — it doesn't fix a process that's poorly designed to begin with. If a workflow is automated before the underlying process is clarified, the result is often a faster version of the same confusion, rather than a genuine improvement. It's worth documenting and simplifying a process before automating it.
Common Mistakes to Avoid
- Automating a process that's rarely used, for the novelty rather than the time saved
- Automating a poorly defined process instead of fixing it first
- Not assigning anyone to monitor or maintain the automation after it's set up
- Underestimating the setup time required to connect systems that weren't designed to work together
The best first automation project is usually the boring, frequent, rule-based task everyone already agrees is a hassle — not the most technically ambitious one.