Several recent authors in philosophy of science—including Weslake (2010), Woodward (2010, 2016), Weatherson (2012), and Franklin-Hall (2016)—argue that the most appropriate description of a particular causal relationship in nature is not necessarily the most detailed or fine-grained description of that trend. My goal in this essay is to provide a methodology for choosing the appropriate level of description for a given causal relationship. Specifically, I argue that choosing the correct level of causal description can be understood as a pragmatic choice that is indexed to a particular decision problem.  I then show that for some decision problems, an agent would not pay any more to learn the value of a more fine-grained causal variable than she would pay to learn the value of a more coarse-grained causal variable.  In these cases, the fine-grained description is not worth anything to the agent, and therefore the more coarse-grained description can be preferred.  Alternatively, if the agent would pay more to learn the value of the more fine-grained causal variable, then she should use the more fine-grained variable in her description.