This paper discusses a number of methodological challenges that ex post evaluation of competition policy faces. We argue that imprecise measurement and simplified analysis not only lead to unreliable results, but also trigger behavioural responses by competition authorities. Because evaluation works as an incentive scheme, it has an immediate impact on the way agencies operate. Faced with simplistic assessment, authorities may be tempted to be overly interventionist, to spend too many resources and to ignore relevant information. Therefore, a careful design of methodologies is important to ensure that ex post control does not bring about the very ills it is supposed to cure.