We have encountered a bug in the pivot grid control that relates to this issue.
When we select 2 measures from an OLAP cube, set against the column area, and a list of items (stores in our case) on the rows, and one of the measures has only empty values, the results returned are completely incorrect. In our example, we select sales and stock for a list of stores (with a time filter set in the report filter). We only need to see the overall results by store on the columns, and the store names in the rows.
If you look at the attached files, the results, in the format we'd expect to see them are shown in the 'incorrect' screenshot, i.e. they all just show the number '6'. I did a SQL Profiler trace on the MDX being generated by the pivot grid and ran it in SSMS manually to see what we get - see MDX query screenshot. Those values were exactly right, but bore no relation to the data shown in the first screenshot.
I did notice however, that the grand totals in the first query were showing correctly, so I changed the pivot grid selection to add a time bucket above the measures in the columns section (so that we can get grand totals on the columns) and the results came back correctly. See respective attachment.
I can see why this is a problem, as the pivot grid is expecting to get back values for two measures, but is only getting one, thanks to the NON EMPTY directive in the MDX query. The problem is that this doesn't currently seem to be allowed.
Can you please raise a support ticket for this issue as this is a bug?