Hi,
It will look for the post tag which matches, the count of items within the taxonomy is not considered in this case. It only checks the text – as fast as possible – and returns immediately if a match is found.
Yes, the ordering matters in this case. It goes from the top, to the bottom of the source list until it finds a keyword from the given source. In your case it will try the post tags first, if no match, then tries the statistics database.