For instance, if a server has 10 or 15 voids in one day, that might indicate a scam. Or if a particular bartender is frequently getting 70 percent tips, that might indicate that he or she is pouring free drinks and being rewarded for it. Each server has a unique login, so a pattern of unusual transactions easily fingers a thief. Additionally, if a certain type of scam is caught running in Chicago, said Lundberg, an algorithm will begin to search for it, and can spot it as soon as it appears in Los Angeles, Miami, or Detroit.