When should you stop an acquisition experiment instead of polishing it again?

Check exposure and measurement first. An undiscovered page or an unpublished creative has not tested demand. Once the intended audience has meaningfully encountered the offer, inspect the stage where progress stops and whether one specific correction could resolve it.

Set a bounded next decision: revise the audience, repair the path, test a clearer offer or stop. Preserve what was learned and avoid endless cosmetic variants. The purpose of a checkpoint is to change the allocation of effort, not to generate another report about the same uncertainty.