Players in smaller clubs will be used to playing "Howell" movements, in which pairs play all or most of the other pairs, sit in different directions on different rounds, and an overall winner is produced at the end of the game.
However, when the field gets larger, it is no longer possible to play all the other pairs because there are too many of them, and it is usual to use a "Mitchell" movement where the NS pairs always play against EW pairs. The trouble with this is that the pairs' results will only be compared with the other pairs in the same line as themselves, and so we have effectively created a "two-winner" movement. Some clubs are happy with this, and produce separate ranking lists for those sitting NS and those who are EW, but many would prefer to have a single winner for their games.
The solution to this has for many years to employ an "arrow-switch", whereby for one or more rounds the players rotate the boards so that the EW players play the NS cards and vice-versa. This means that you will compare some boards with those sitting in the opposite direction as well as those sitting in the same direction as you, and so a single-winner ranking list can be produced.
One of the interesting and perhaps counter-intuitive things about this is that it's not necessarily better to arrow-switch more rounds. Detailed analysis of arrow-switching by JR Manning in the 1970s concluded that the ideal number of rounds to be arrow-switched is about one-eighth. Switch much more than that, and it becomes self-defeating. In practice most clubs switch two rounds when they are playing two-board rounds and one round when they are playing three-board or longer rounds. It's common to arrow-switch the rounds at the end of the game, but there's no reason why it shouldn't be at the beginning.
Readers who are interested in this subject can find Manning's paper at
http://www.ebu.co.uk/documents/laws-and-ethics/articles/bridge-movements.pdf