Tableau's box plots are convenient and attractive, but require fetching all the data points to Tableau that will influence the box plot, doing all the percentile computations on the client instead of at the data source. This approach doesn't scale to large data sets. This simple example shows how to leave the data at the source and still draw a box plot efficiently. It relies on the Percentile() function which Tableau only implements for some data sources :-(, but there are alternatives described below. By contrast, histograms efficiently leave the data at the source and only fetch back the counts for each bin.