Where do you start with automation?
Not with the biggest task, but with the most predictable one. Four criteria point to the right candidate.
It repeats
A weekly task is worth more than an annual one, even if the annual one takes longer.
The rules are clear
If you can explain the task in ten sentences without "it depends", it can be automated. If you cannot, start by clarifying the process.
Nobody likes doing it
That is a serious criterion. The work everybody postpones is where mistakes accumulate.
A mistake is visible
Automate first what you can check the result of. Invisible automation that silently gets it wrong is worse than manual work.
What you measure first
How long the task takes today. Without that number you will not know whether you gained anything — and you will not be able to justify the next step.
