This paper addresses the question, what metrics should be used for performance evaluation and in particular how they should be weighted and combined in the presence of technological interdependencies when the agents exhibit variedly strong developed rivalry. We find that the principal reacts to his agents' competitive preferences through a reallocation of incentive intensity. As a consequence, depending on the underlying sort of technological interdependency, various differences in the balancing of performance measures compared to the case of purely egoistical behavior arise and changes in the agents' basic types of compensation can occur. We further show that the principal does not want both of his agents to behave equally competitively. Instead, he can only profit when the agents are asymmetrical. Then the principal wants the more productive agent to exhibit rivalry while the other ideally should behave completely egoistically.