It doesn't work well for highly precise values, or small datasets, or prefixes. Which are our three use cases.