Instead of using a filter, a parameter and calculated field could be used to see which customers had sales and which did not. I have attached a sample workbook with instructions. Hope this helps!
sales in period.twbx.zip 1.1 MB
Thanks for your note. Could you send the workbook in Tableau 6.1? I haven't migrated to version 7 yet.
Thanks again for your help and I look forward to seeing what you did with your workbook.
Piggybacking on Omar's post (I'm working with him on the same project): what's happening with this data is that the datasource is updated with daily values for businesses that have had transactions, but only for those businesses. If a business has no activity for that day, it doesn't appear in the record at all, so there are not even nulls to work with. Now, if the business has at least one entry in the timespan being viewed, calculations like the ones used in http://community.tableau.com/thread/116429 will convert the missing data to nulls and then to zeros, but if there are no transactions the business drops out of the view entirely.
Reading up on other cases where missing data needed to appear or be replaced, I saw a join-with-default-values strategy being mentioned in earlier threads:
This looks like it could solve our problem, although I'm not sure as to what should be in the source used to pad the data. The second thread contains examples, but looking at the workbooks I can only find the primary, sparse data, and not the values used for padding. (Note that I am very new to Tableau, and so could easily be overlooking the data in question). Are they just simple lists of the possible dates and other information being extracted, with the SQL both creating the combinations and making the joins (my SQL is also pretty basic)?