I figured something out, though I would have liked the normal Xref documentation as Rahul Singhal pointed out to have covered this.
So to be clear, the normal Xref documentation works to get a cross reference, but with the title instead of the (in my experience common) type and number reference, e.g. Figure 5.6 / Table 4.
The solution I found still uses Xref, but the inspiration came from 'Number Section Headings, Table Titles, and Figure Captions Programmatically' documentation. Number Section Headings, Table Titles, and Figure Captions.
I modified the getFigureCaption and getTableTitle functions shown in the above documentation. So instead of using the Dom functions to caption figures (linkTarget on fig.Snapshot.Caption) and tables (linkTarget on BaseTable('Title', ...) I build my own paragraph with my own counters and set linkTarget to the Figure M.N portion / Table M.N of my captions. A bit of work but it does work out.
