Over-time variation in individual’s customer satisfaction scores

Abstract

Customer satisfaction is a commonly used business performance metric. Despite the widespread use of satisfaction surveys, little is known about how stable individual’s satisfaction scores are. If individual’s scores show instability, this has implications for market research design and managerial actions. To investigate the stability of satisfaction scores, this study uses data from a two-wave satisfaction survey in which the same respondents were interviewed 6 weeks apart. The respondents had no recorded purchase with the retailer between survey waves. The main finding is that only 49% of respondents give exactly the same satisfaction score on a 1–7 scale when re-surveyed. After aggregating the results into three simple categories of dissatisfied, neutral to somewhat satisfied, and satisfied, the proportions who stay in the same category from one survey to the next are 44%, 57%, and 82%, respectively, despite the overall average score for the sample staying the same. The changes in scores are a manifestation of regression to the mean, whereby those who give a low or a high score the first time tend to regress up or down toward the overall average score the next time. The main management implications are (1) interventions aimed at low or high-satisfaction customer groups need to take regression to the mean into account; (2) attempts to relate individual’s satisfaction scores to future behavior (e.g., loyalty, brand switching) should use scores averaged over two surveys; and (3) the oft-quoted belief that dissatisfied customers will tell more people compared with satisfied customers is less tenable, given that low-satisfaction scores tend to regress upward more than high scores regress down.