The relationship between IT investment and performance improvement has been a classic research subject. In earlier research it was characterized as the famous “IT productivity paradox.” Decades of research effort later, many IS/IT researchers no longer consider this relationship paradoxical, but it remains a challenge to fully explain “IT productivity variance,” that is, a large portion of unexplainable performance variance observed in empirical studies. In this paper, we take a theory development approach to this important issue. Our effort focuses on examining the theoretical foundation of “IT productivity variance.” Through literature review, we identify and contrast three prevailing theoretical perspectives to suggest the plausibility of a contingency approach. Contextual factors can significantly affect the relationship between IT investment and performance improvement. We identified three such factors that were not thoroughly examined: level of performance measurement, technology type, and technology life cycle. We subsequently propose a contingency model in light of these factors, and develop a set of research propositions. Our effort is a first-step toward bridging a major gap in the literature. The proposed model, while awaiting empirical validation, can provide useful guidance to practicing managers.