Canary Categories

Past customer spending in a category is generally a positive signal of future customer spending. We show that there exist “Canary Categories” for which the reverse is true. The more customers purchase in these categories, the less likely these customers are to return to that retailer. The finding is robust and replicates at different retailers.

We show that the effect results at least in part from heterogeneity in preferences: different customers prefer different brands. Because retailers cannot stock every brand, some customers cannot always find their favorite brand. While these customers may purchase on the current trip, some may not return to the retailer on future purchase occasions. We support this explanation with evidence from both observational field data and laboratory experiments.

The findings have important implications for retailers determining retail assortments. Existing best-practices recognize the role of brand substitution in assortment decisions, and argue for using temporary stockouts to measure brand switching. In canary categories, retailers also need to consider store substitution and temporary stockouts may under-estimate store substitution. In-store assortment experiments can overcome this limitation, but these experiments must be carefully designed to measure both brand and store switching.