To clarify, loosely speaking, it could be something like this:
For a single month, two peoples entries A1 and B1 coming
first and second might get 10 votes and 7 votes from 20 voters. Normalisation would yield 10/20 and 7/20, giving 0.5 and 0.35. With a simple scaling factor, say 10, the score on the scoreboard might then be 5 and 3.5.
For a later month, the same two peoples entries A2 and B2 coming
second and first may get 6 and 10 from 17 votes. Normalisation yields 6/17 and 10/17, giving 0.35 and 0.59. With the same scaling factor giving 3.5 and 5.9. Added together would be 8.5 and 9.4. The person entering shots B1 and B2 therefore wins because, despite the reduced number of voters in the second stage, the proportion of 10 to 17 is better than 10 to 20.
This is just an example of my point, and I'm not suggesting this is precisely the way it's done, but you see what I'm getting at