Purpose – While satisfaction, value, image, and credibility are commonly assumed to drive customer loyalty, there is nevertheless reason to question whether their effects vary across groups of consumers. This paper seeks to explore how individuals with contrasting need-for-cognition (NFC) levels differ in using memory-based information when forming behavioral intentions towards a current service provider. Design/methodology/approach – The authors tested the hypotheses by means of survey data from customers of retail banks, and applied two-group analysis using structural equation modeling (SEM) to test the moderating effects of NFC. Findings – Satisfaction positively affects loyalty for high NFCs, but not for low NFCs. Image is insignificant in both groups. Value positively affects loyalty for low NFCs, but not for high NFCs. Credibility has a positive effect for low NFCs, but not for high NFCs. Research limitations/implications – The limited sample size affects the power of the test methodology, but Chow-tests of regression models gave similar results. Further research should test the model in other contexts to enhance external validity. Practical Implications – To develop more effective customer strategies, both researchers and practitioners need to understand how different types of consumers attend to and utilize information when forming behavioral intentions. The standard practice of surveying customer satisfaction and loyalty typically requires the consumer to make a memory-based judgment. Originality/value – Previous research has primarily focused on how consumer demographics moderate satisfaction-loyalty links. This paper includes additional drivers of loyalty, and assesses moderation by a personality trait (NFC) not previously used in satisfaction-loyalty research. The results indicate that a consumer ' s dispositional tendency to think and elaborate (more or less) can bias survey results.