Why should a project define a success metric before comparing models?

A. It determines the programming language automatically B. It removes the need for test data C. It guarantees that every model will perform equally D. It creates an objective criterion aligned with the project goal