This paper introduces and discusses a mechanism which minimizes gaming or manipulation activities, if payments are linked to results from manipulable methods. The idea is to add non-manipulable information to manipulable information to improve the evaluation of a given output. A score declining in increasing evaluation quality indicates gaming. A simple agency problem explains the advantages and disadvantages of the approach. The introduced mechanism dominates a single evaluation method if risk aversion is positive but not too great. The principal should however let each agent decide about which evaluation method he prefers. The mechanism is applied to performance measurement in research organizations and hospitals.