Is last-mile delivery only viable in densely populated centres? A preliminary cost-to-serve simulation for online grocery in the UK

Abstract

This paper proposes a conceptualisation of the ‘cost-to-serve’ linking the economic viability of last-mile delivery. Specifically, a cost-to-serve model is developed against business-specific indicators, such as share of online sales for a given catchment/geography. The outputs are then discussed in terms of e-commerce penetration and retailer density for the designated catchment/geography. Preliminary evidence is generated by means of computer simulation for a case in online retail of fast moving consumer goods (FMCG) with a focus on the UK. To this purpose a three-stage, pragmatic simulation approach is outlined, using widely available data in order to evaluate alternative last-mile network configurations, and their associated cost-to-serve cost curves. At each stage a sub-model generates, respectively, 1) order-generating locations; 2) basket composition; and 3) last mile delivery cost. Off-the-shelf tools are employed throughout to generate and visualise key analytics, thus facilitating replicability in real-world industrial settings. Results suggest that as well as having cost benefits with increased market penetration and/or increasing the number of drops per journey, as might be predicted, there appears significant potential to narrow the spread of cost variability for a given market penetration by influencing the available locations/time/price options to online customers. The proposed approach can support similar developments besides FMCG, for example in the pharmaceutical industry as direct-to-home medicine delivery becomes a credible option.