When does a manual workflow become worth automating?

How do you decide when a manual workflow has earned a productized version?

I’m building IndieNeed, a directory for indie products, and I’m intentionally doing a few things by hand: reviewing submissions, cleaning up descriptions, and checking whether categories make sense. That keeps quality consistent, but it also creates a queue and makes me wonder when a repeatable rule should become code.

For founders running directories, marketplaces, or other data-heavy products, what signal tells you it’s time to automate? Is it volume, repeatability, error rate, or a specific kind of user frustration? I’d especially love to hear about automations you delayed on purpose—and what changed your mind.

The current shape of that workflow is visible on IndieNeed’s simple submission page: https://www.indieneed.com/submit