Attitude vs Action - The predictability of attitudinal estimation models on purchasing behaviour

Sammanfattning: Understanding consumer behaviour has become increasingly important for decision makers. As a result, tracking data to assess consumer attitude, intention and loyalty is growing in popularity. While survey based insights provide knowledge about intended and probable behaviour, this approach does not have full descriptive power for actual behaviour. Attitudinal estimation models aimed to describe purchasing behaviour have emerged. These models, however, are influenced by exogenous factors in the form of measurement errors and systematic biases, leading to discrepancies between attitude and action. This thesis tests the discrepancy by comparing attitude and action. This is achieved by defining customer share of wallet from the Wallet Allocation Rule and putting this measure in relation to reported purchases for the same customers. To determine what exogenous factors influence the behaviour, both pre-purchase estimates, in the form of purchase intention, and post-purchase estimates, in the form of loyalty scores, are applied. The results show a clear mean difference between attitude and action in all investigated product groups. In addition, it is emphasised how pre-purchase estimates are influenced by economic and in-store variables, such as price, availability and promotional offers. On the same note, post-purchase estimates are influenced by the quality of the product and by cognitive factors, such as brand recognition, preference and trust. These insights are complemented with category specific influencers to provide managers with knowledge on how to increase wallet share, at an individual level, and market share, at an aggregate level.