Automation delivers value when it is applied to the right work. Applied to the wrong work, it adds fragility and maintenance for very little return.
A short scoring exercise usually settles the debate faster than a long discussion.
Score each candidate process
High frequency, high effort, stable and clear processes are strong candidates. Anything with severe consequences needs human review built into the design rather than exclusion from automation altogether.
- Frequency: how often does this run in a month?
- Effort: how long does one instance take a person?
- Stability: how often do the rules change?
- Clarity: can the decision logic be written down?
- Consequence: what happens if it is done incorrectly?
Simplify before automating
Many processes contain steps that exist only because of a former system or a former colleague. Automating those steps preserves them indefinitely.
Remove what is unnecessary first. The remaining process is usually smaller, cheaper to automate and easier to maintain.
Design the exceptions
The success of an automated workflow is largely determined by how it handles the cases it cannot process. Decide early where exceptions go, who reviews them and how quickly.
Without that, the first unusual case becomes the reason the team stops trusting the workflow.