I think I've got a workaround, but it's not seed data, which would upset functions like count. Programmatically construct a range of intended pivotitems, build a p-cache on it, then change the SOURCEDATA of the cache to the report data and refresh the table. I've got a working prototype. I'm...