Any way to identify the contents of each dataset column in the "Select column" listing?

I've just been drafting the first of our testing scripts and have found it impossible to know the type of data contained inside each of the columns in the _Recycling Banks_ dataset. It's unclear whether the column titled "Batteries" (for instance) contains data detailing the _Number of batteries recycled_ or _Type of batteries recycled_ or something else entirely...

Of course, if I was viewing the original data table with all the data in rows under the column heading, then more detailed labels would probably not be needed. But users of COD will likely never actually see the original data table. So without a preview on the _Select column_ listing, they have to guess, submit, check it's what they wanted, then go back if it's not what they were expecting. Since we don't have a live preview of the visualisation on the page to help users discover if they've chosen the right column in situ, perhaps a way to help might be to somehow auto-detect the type of data inside a column and display either an icon (indicating whether it's an integer, time, date, plain text, etc) or even a short preview of a few rows of the column contents.

BTW, if this isn't a @pmackay issue, please reassign.