Great callout. The paper cites exponential for revenue data, but yes you can use any continuous distribution. I see how the blog makes it seem like you only use exponential - thanks for clarifying in this comment!