So this was, of course, a fictional story, but situations like this seem to happen again and again.
I've attached the sample data as well as three QVWs that represent the three examples above. QlikView is an amazing tool and can do amazing things, but it is very difficult to explain to customers when something seemingly as simple as this causes issues.
For this particular Rolling 13 month scenario I have tried all of the following:
An expression for the Date Dimension that limits it to the past 13 months
Every combination of "Suppress When Value Is Null" and "Show All Values" that I can think of on both Dimensions
Every combination of "Suppres Zero-Values", "Suppress Missing" and "Populate Missing Cells" I can think of
ValueList based dynamic Dimensions for Date
Simple if..then expressions for the sales amount
Set Analysis based expressions for the sales amount
Individual expressions for each month (works on the table, not on the chart)
The only thing I haven't tried, and seems like a real hack, would be to populate the data with a dummy record for each SalesPerson for each month with zero value so there is at least one entry each month for each person. I may try it just to see if it works, but in the real situation, where there are millions of rows in the source data, this would be very unpalatable.
I hope I am missing something obvious here, and I would be ecstatic to be told as much.