Which close tasks are worth automating first?

Detailed Explanation

The repetitive, rule-based ones with a clear right answer: accruals that follow a formula, intercompany matching, and variance analysis that compares against a defined threshold. These are good candidates because the logic is stable, the volume is high, and consistency is itself a benefit — an automated accrual is calculated the same way every month, which removes a source of period-to-period noise. Judgement-heavy tasks are poor first candidates. Automating a decision nobody has articulated tends to encode whichever version the last person used.