A project lead turning a proof of concept into a governed process is deciding which transformed modules should go through an optimization pass first. Which is the appropriate basis for that decision?

The correct basis is choosing modules whose clarity or maintainability can be genuinely improved without altering their validated required function. Optimization is targeted work, applied where it improves clarity or maintainability while the required function stays fixed and already validated. Choosing whichever modules were transformed most recently, regardless of their current readability, is incorrect because recency says nothing about whether a module needs or benefits from optimization. Choosing modules where a rewrite could also change their business behaviour to something considered more modern is incorrect because changing behaviour during optimization reopens equivalence questions transformation already closed. Choosing all modules at once, applying the same optimization uniformly without review, is incorrect because uniform, unreviewed changes risk regressions across modules that did not need touching. For the exam, base optimization decisions on maintainability benefit, with function held constant.