Loading
Leave it alone. Automating judgement produces confident mistakes at speed.
Weighted towards frequency and whether the rule can be written, because those two decide it. The rest describe how safe it is to try.
It helps you list the manual steps repeated across your teams and assess each for whether it should be automated — including the ones that should not.
The assessment matters as much as the list. Automating a step where somebody was quietly checking something is how silent errors start.
List steps that happen more than weekly and follow the same shape each time. For each, note who does it, how long it takes, and what they check while doing it.
That last column is the one that decides. A purely mechanical step is a candidate; one with a glance attached is not.
It will not tell you the automation will work. Some steps are repetitive because the underlying process is broken, and automating them makes the breakage permanent.
It also cannot see the checks people do without realising.
Rank by total time, then filter out anything involving judgement, anything customer-facing without review, and anything whose failure would be silent.
What remains is a genuine backlog. It is usually shorter than the original list, which is the point.
Judgement, anything customer-facing unreviewed, and anything failing silently.
Ask the person what they look at while doing it.
Yes.
Half an hour on your own figures, and an honest answer about the parts Treepie does not improve.