> Part of this is definitional, too. Do you include that far left part of the graph where people are not really paying you as a "customer"? Exactly. Often when

> Part of this is definitional, too. Do you include that far left part of the graph where people are not really paying you as a "customer"? Exactly. Often when the mean is problematic, it's because there's multiple subgroups in the data with very different behavior, and when you split those up, you can learn a lot more from the numbers.