Purpose – Customer loyalty is a focal concern for marketers who seek to identify its antecedents and causal structure with the aim of better understanding, predicting and managing loyalty. The purpose of this paper is to model both current behaviour (measured as share of wallet) and future intentions as measures of customer loyalty, to quantify the link between current and future behaviour. Design/methodology/approach – A hybrid model, combining reflective and formative constructs, was developed, moving away from the traditional “reflective only” approach to explain customer loyalty. New predictors such as variety seeking, “resistance to change” and risk taking behaviour were tested to explain loyalty. Findings – While “risk” is traditionally viewed as a key variable in financial services, this study finds that variety seeking and “resistance to change” predicted current behaviour and future behavioural intentions better than risk. Higher explanatory power and better model fit was found for a hybrid model combining formative and reflective constructs; in contrast to the more common fully reflective approach. Research limitations/implications – This study adds to the emerging debate on whether concepts such as loyalty should be treated as reflective and/or formative. The implications from this study suggest that future research can usefully model current behaviour as formative and future intentions as reflective. Future research should test the extent that these findings apply across products and services beyond banking. Originality/value – This study establishes that variety seeking and “resistance to change” can usefully explain and predict loyalty. The examination of “formative” and “reflective” concepts in explaining loyalty is also novel.