: The HCI community has long been accused of delivering ‘common sense’, ‘useless’ information, and to be ignorant of business needs. HCI experts are also criticized for failing to provide or apply theory-based techniques. This paper shows that the two goals may be incompatible. It discusses one case study in which HCI data intended for one purpose were inappropriately applied to support another. Theory-driven GOMS (Goals, Operators, Methods, Selection rules) models generated to predict performance in two competing applications were subsequently used for making a business decision. Three similar data-driven studies designed to inform a business decision are then presented. Findings from all these studies demonstrate that the parameters on which the business decision based on GOMS data was made were largely irrelevant to that decision. It is argued that HCI experts must learn to relate their findings to business needs and values if HCI practice is to progress.