The three business tasks worth automating first
Where automation pays back quickly, and where it usually is not worth the effort.
Automation is worth doing where the rules are already known and the task repeats. That narrows it considerably.
The first candidate is follow-up. Quotes that go unanswered, enquiries that arrive after hours, appointment reminders. Nobody enjoys chasing, everybody forgets, and the rules are simple.
The second is data re-entry. Anywhere a person copies information from one system into another, an integration can do the same job without transcription errors.
The third is reporting. If someone spends a morning each week exporting and reconciling spreadsheets, that morning is recoverable almost entirely.
What is usually not worth automating first: judgement calls, exceptions, and anything where the rules change every month. Those are better handled by making the information easier to find, so the person deciding can decide faster.
The honest test is simple. Write down the steps. If you cannot write the rule, the software cannot follow it either.