How to Analyse Shopping Data for Social Research: Urban retail patterns

Are you keen on understanding the intricate dynamics of consumer behaviour and urban retail patterns using shopping data? The use of loyalty card data has emerged as a powerful tool in social research, offering insights into consumer habits and preferences. However, the process of analysing this rich data source involves meticulous methods to ensure accuracy and meaningful interpretations. Let's delve into the methods used by Rains and Longley of analysing shopping data for social research.

Understanding the Geography of Consumption

The first step in analysing loyalty card data is to comprehend how store networks shape consumer behaviour regionally and locally. Loyalty cards are widely used across the UK, potentially representing a diverse demographic. However, the geographical unevenness in store coverage needs careful consideration in the analysis process. This involves controlling for variables like the distance travelled to a store or purchase frequencies, which can distort geographical measures​​.

Defining 'Complete' Cardholder Records

A crucial aspect of analysing loyalty card data is defining what constitutes a 'complete' cardholder record. Since transactions without a loyalty card swipe go unrecorded, determining completeness involves comparing the breadth of purchases across various product categories with the consumption domains of the Office for National Statistics' Living Costs and Food Survey (LCFS). This method helps in overcoming the limitation of lacking information on purchases made outside the retailer, thereby providing a more holistic view of a household's purchasing habits​​.

Initial Data Processing and Completeness Scoring

The initial data processing includes matching transaction data to the LCFS COICOP scheme. The mean number of categories purchased weekly is calculated, and these ranges are used to categorise the weekly purchasing for each household in the loyalty card data. Completeness scoring is based on 'normal' purchasing within the week, rather than the RFM approaches which often favour higher spenders even if their spending is limited to a single category​​.

Periodic Scoring and Segment Analysis

The analysis in this particular instance extended to aggregating weekly scores into a total score for each 13-week period. Households are then categorised into segments, reflecting their loyalty and purchasing patterns. This method provides a nuanced understanding of how often and how much households engage with the retailer, highlighting patterns of regularity or sporadic purchases​​.

Geographical Coverage and Consumer Behaviour Insights

Further, the analysis identifies the uneven geography of coverage within the 'complete' segment. For instance, it reveals differences in shopping habits among various demographic groups, like Rural Residents, Suburbanites, Urbanites, and Cosmopolitans, reflecting their respective lifestyles and consumer cultures​​.

Creating Synthetic Estimates for Uncovered Areas

In areas with little or no loyalty card data coverage, synthetic estimates are created, 'borrowing strength' from areas with higher coverage. This method allows for a comprehensive map of estimated neighbourhood interpurchase periods, providing valuable insights into how different communities approach shopping​​.

Triangulation and Validation with Traditional Sources

The value of consumer data from loyalty cards, although immense, requires careful triangulation and validation with traditional data sources like surveys and censuses. This ensures that the data's self-selecting nature does not skew the research findings and that interpretations are grounded in reality​​.

Conclusion: Towards a Critical and Exploratory Use of Loyalty Card Data

Researchers must approach loyalty card data with a critical and exploratory mindset. The potential of this data is immense in understanding consumption differences across population segments. However, it is imperative to recognise the limitations and ensure that the data is used in a way that is descriptive and grounded in established sources for purchasing patterns​​.

By employing these methods, researchers can unlock the full potential of shopping data for social research, paving the way for innovative and comprehensive studies in consumer behaviour and urban retail analytics.

Source - Rains, T. and Longley, P., 2021. The provenance of loyalty card data for urban and retail analytics. Journal of Retailing and Consumer Services, 63, p.102650.