A flow that works in a test environment doesn't automatically work in production either. That sounds obvious, but it's the most common cause of automation incidents we see with new clients.
Where it goes wrong
In a test environment, edge cases are scarce: a handful of test contacts, a handful of orders. In production, edge cases show up on their own: a customer without an email address, an order with a negative amount due to a return, a webhook that arrives twice within a second. A flow that hasn't been explicitly tested for that gets stuck in production — usually at the worst possible moment.
Recovery is rarely quick. It involves unwinding incorrectly sent emails, correcting faulty data across multiple systems at once, and explaining to a client exactly what went wrong.
What a good test round actually does
Test with real edge cases, not just the happy path.
Simulate duplicate and delayed events, not just the normal order.
Have someone other than the builder design the test scenarios.
That last step sounds small, but is often the difference: whoever built a flow unconsciously tends to mostly test the path they already thought of.