The tool is estimating the reach as a number of matching keywords between the text of a mention and found keywords. 

It can happen that much more keywords are found in a mention inside one query than in the other, thus such large difference in reach.

Did this answer your question?