The provenance of loyalty card data for urban and retail analytics

The deployment of loyalty card and other consumer data in geographic research brings opportunities to explore and understand patterns of purchasing behaviour in unprecedented detail. However, valid generalisation requires thorough evaluation of their potential bias. We argue that, in competitive markets where consumers can choose to shop across competing retailers, loyalty card data from just one of these may not represent a ‘complete’ view of all purchases, and that this ‘completeness’ must be controlled for when assessing bias. To this end, we undertake a UK wide analysis of loyalty card data assembled by a major UK grocery retailer and provide guidelines for their effective deployment in the domains of urban and retail analysis. We assess, for the first time, the ‘completeness’ of circa 500 million customer transactions recorded by a major customer loyalty programme in representing the overall purchasing patterns of circa 16 million consumers across the entire UK, and develop a method by which to do this. Moreover, no operator has complete national store coverage, and so we suggest ways of accommodating this when conducting analysis using loyalty card data. We illustrate the importance of these issues before providing recommendations for the wider use of consumer loyalty card data.