Job attitudes and performance are correlated, but which comes first? Despite a long-lasting debate regarding this question, the potential ordering of the job attitude–job performance relationship and its ideal timing of measurement over time remains unclear. Based on the hedonic treadmill theory and the endowment/contrast model, we develop arguments as to why the influence of performance on subsequent attitudes might be less persistent over time than vice versa and we suggest a strategy to determine the ideal period for measurement time lags. We contrasted both temporal directions within a data set of 755 employees from the retail banking division of a large bank, nested in 34 business units, for the period of 2005–2008, allowing for a controlled environment and consistent data capturing over time. We studied the relationship of organizational commitment aggregated to the business unit level with two business unit performance indicators (financial achievement and customer satisfaction). Results indicated that organizational commitment had a more persistent influence on performance at the business unit level than vice versa. Consistent with prior research, this suggests that job attitudes may come first, and that practitioners might be well advised to aim to improve job attitudes in order to boost performance.