I find this a little hard to parse because it's done on a percentage basis -- ex, many high value users stayed high value, but this diagram makes it look like all high value users either became inactive or low-value. Seems like we should also add high-value as a bucket on the right side of the graph for the users who stayed?

You are totally right! Sankey's struggle with loops. I'm gonna have to put high-value -> high-value on both sides to make it make sense. thanks! (also i didn't mean to resolve your comment! i pressed the wrong button. Am new here.

Ok, the charts make more sense now! I'm going to fix up the colors so each state has a stable color which will help make them scanable

Thank you!!