During the Advanced Extractions presentation at TLCC2015 there was a question or two about how the randomization takes place for the Random Extract and Random Select functionality. I wanted to clarify how this works since I, in retrospect, may have only partially answered the question when asked.
Both the Random Extract and Random Select features use the NEWID() function which creates a new and random global unique identifier (GUID) on each row of assigned customers in a segment. From there, the assigned customers are sorted by that new GUID, then a TOP ## PERCENT SELECT is used to grab the ## percentage needed.
If you are using Random Select, the percentage of customers left over are available to be selected in future extraction segments. If you are using Random Extract, they are not.