Robust statistical methods for point of sale designs, the example of healthcare snack and drink vending

Purpose: The experiment introduces nutritional labelling, healthier products and product placement designs to the hospital vending machines, to promote healthy lifestyles. / Design/methodology/approach: The site where this experiment happens is a major London hospital, serving over a million patients every year. In the experiment, the hospital’s snack and drink vending machines are redesigned. The impact on product sales are then analysed using robust statistical methods. / Findings: Nutritional labelling has a statistically significant impact on product sales. Less of the unhealthiest products are sold. Healthier products and product placement designs have a larger impact but with less statistical significance. They require further testing. / Research limitations/implications: Experts in service operations can use this experiment’s regression modelling methods. The methods are ideal for measuring change over time in counting data sets in complex real world environments. / Practical implications: There are suggestions for practical vending service change in this research. They are in line and add a practical example to Government policy guidance. / Social implications: People using the redesigned vending machines have more opportunity for healthy lifestyle choices. / Originality/value: The experiment provides statistical evidence in support of catering for healthier lifestyles