How Do You Know When a Trading Loss Is Useful Information Rather Than Just a Normal Outcome?

A losing trade does not automatically mean the strategy or execution was faulty. Sometimes the setup simply failed within the expected range of outcomes. How do you distinguish an ordinary loss from one that reveals a problem worth addressing? What evidence do you look for before changing your process?