An example of one that disappoints is https://www.reddit.com/r/dataisbeautiful/comments/15f01pb/oc... where that one big bucket of 'total applications' loses a lot of data. Was everything from Dice ghosted? That would be useful information. I would contend that there's no value in the 'total applications' bucket in there at all other than to sum up the leftmost column - which can be done separately.
One big bucket isn't necessarily wrong, but if one is trying to show off the features of the Sankey diagram, showing it with the stronger representations that it is capable of doing can be useful.
Having seen countless poorly done ones with an everything bucket in the middle that squashes valuable relationships between the inputs and outputs... and that's a turn off for me when they are presented that way.
No comments yet.
