The importance-performance analysis (IPA) is widely used for identifying which quality attributes should be improved for maximizing user satisfaction. The two dimensional grid of IPA are based on user-perceived attribute importance and performance. If the user-perceived attribute importance is high but its performance low then enhancing the performance of this attribute is likely to result in higher user satisfaction. But some studies have found that user importance may depend on attribute performance. This confounds the IPA analysis. Yet there is no study which has investigated whether this phenomenon is applicable for IS (Information Systems) products. This study conducted with an ERP system users show that user importance of an attribute is indeed dependent on its performance. For some attributes users overestimate their importance when the performance is low and underestimate them when the performance is high while for others the reverse is the case. Implications of this phenomenon for practice are discussed.