The point of it
The idea is to break down your filtered results by the values found in headers in the corpus. For example, if your corpus has a <Newspaper>...</Newspaper> header, the procedure will find the string between those two tags and compute sort scores based on those strings.
How to do it
With a set of relevance results on screen, press the Breakdown button. Type in the header text.
The program will offer to copy results to Excel, or just copy them to the clipboard.
Something like this.