Refining spatial grocery models of consumer behaviour: an individual-based approach

Consumer grocery behaviours have evolved over the past few decades, with consumer transaction behaviours becoming increasingly individualised due to the changes in cultural lifestyles, the increased supply and demand for convenience, and the expanded provision of e-commerce. The UK grocery sector is a competitive market, and optimising their store network is of most importance to retailers. Current tools used within grocery location analytics use a top-down methodology, such as spatial interaction models. In the past, these tools have been considered the most appropriate for location impact assessments and predicting new store revenue. Whilst robust for the time, these models do not necessarily account for one of the most integral parts of store performance, the consumer's behaviours. Due to their top-down nature, current spatial models cannot account for the heterogeneous behaviours of grocery consumers, which are becoming increasingly diverse. These models would benefit from refinement in their ability to capture the nuanced behaviours of customers.

Therefore, the research presented in this thesis provides a framework for developing an individual-based model using modelling elements from decision trees, microsimulation, and, largely, agent-based modelling. First, a rarely accessed loyalty card-linked transaction dataset provided by the study collaborator, Sainsbury's, was analysed and segmented, identifying seven unique consumer type groups. Using these groups, a fully reproducible individual-based model was developed. The study highlights the advantages these models could bring to retailers by modelling transaction temporality, and the notable challenges encountered when modelling spatiality. The research in this thesis provides a novel framework to create such data-driven bottom-up models that utilise transaction data by known customers. The results of this thesis present how individual-based models could be developed to provide grocery retailers with an integral tool that captures the diverse behaviours of consumers for site location analysis via scenario testing.