This chapter investigates the question whether predictive policing actually prevents the occurrence of crime. In light of the political prioritization of burglary prevention and the public attention that predictive policing was subjected to, police departments needed to demonstrate that what they were doing was in fact successful. Coming up with proof for the success of targeted crime prevention is, however, not easy. Complex social settings with dynamic interaction effects, as well as numerous possible intervening variables, render it almost impossible to attribute the nonoccurrence of crime to the use of predictive policing software in a causal fashion. When faced with these challenges, evaluation studies of predictive policing thus opted to redefine success criteria instead. Rather than aiming to establish statistical evidence for the assumed relation between algorithmic crime analysis, operational measures, and decreasing residential burglary numbers, police departments opted to highlight long-term organizational benefits of the use of crime prediction software. By foregrounding aspects such as improved data-handling capacities and communication processes, success was framed in a sociotechnical fashion as the capability to produce and transmit knowledge and power across different parts of police work and to bring risk from the analyst’s desk to the streets.