If you then add an autofilter to a pivot table the way mentioned above, then here’s what you see:

The option that used to read ‘More sort options’ now reads ‘Sort by Color’….although it in fact allows you to click on the ‘Custom Sort’ button and sort by something actually useful.

This is a neat trick, and one that I’ve resorted to when the calculated fields didn’t calculate what I wanted.

You know how to get the original data out of a pivot table, don’t you? Double clicking on any cell in a pivot table inserts a new sheet with the records used to calculate that cell. Double clicking on the grand total cell gives you all records in the original source data.

As I point out above, a problem with using a filter down the side of a pivottable to filter it can cause some issues if you are also using filters in the pivot fields. And as I also mentioned above, I’ve previously been aggregating up some of my pivot table data fields into aggregate totals using a bit of SQL, so that they could then be dropped in to the pivottable as row fields and filtered.

But today, I tried a different tack, after realising a pivottable can be used as a datasource for another pivot table.

One bonus of this method is that you can effectively add ad-hoc fields to your original (Master) pivottable, that then show up fully intergrated in the new (Slave) pivottable. That is, you can build some ad-hoc formulas that far surpass what a calculated field will allow you to do, or introduce new data altogether (perhaps from a seperate data source, as I was doing today). This can then be dragged and dropped into the new pivot table as row fields or data fields, just like the original data can. So you can happily filter away, as well as sort the Slave pivot table by the new fields (which you can’t do with the autofilter hack above).

For instance, if you want your dates to coincide with a fiscal year that runs from 1 July, but you can’t get at the original data source to add another field, or you want a quick and dirty workaround, then you could put a new column to the left of your original pivottable (rather than the right, where it would get overwritten if the Master Pivot Table has any more row or data fields added to it) and use the formula to return a financial year field, using something like:

=IF(MONTH(B2)>6,YEAR(B2)+1,YEAR(B12)) …assuming this formula is in A2 and copied down, and the pivottable has dates in row B2, and you put ‘financial year’ as a header in A1

Then when you selet the data source for the Slave pivot table, just be sure to include row A through to the last row that has pivot table data, and you’ve solved your problem. Probably best to define a dynamic range for this in case the Master pivot table grows or shrinks.

This might be a convenient way to bring new data into a pivotchart too…if pivot charts are your thing.

What do you think? Worthwhile? Drawbacks?