Europe’s second-largest airline, flying more than 70 million customers across 800 routes between 130 airports.
Reduced margins on ticket prices make secondary revenue streams even more important to low-cost carriers. While many airlines would like to expand their in-flight shopping catalogue, weight and size constraints make that difficult. As a result, the airline was looking to have one catalogue to serve the shopping needs of customers flying across all 800 of its different routes.
Trolleys are loaded in one of 27 different bases, and during the course of a day each trolley will fly more than one route, encountering more than one type of customer. Routes where passengers are predominantly holidaymakers are among the most valuable, while on other routes sales are so low as to be almost negligible.
Faculty was asked to develop a model that would optimise in-flight shopping trolleys for maximum value according to how heavy they were and the types of passengers likely to be on the plane.
Using the knapsack optimisation algorithm, we created a model that would take into account weight and volume constraints. We analysed past sales data to gain a clear picture of the different shopping patterns of passengers across different routes. The model took into account past spending behaviour to estimate how likely it was that each product would be sold on each route during the day, and assigned relevant values to each product.
Looking at data from 2016, we estimated that trolleys optimised in the way we proposed – and carrying the same number of items – would have increased the airline’s revenue by tens of thousands of pounds.
As well as decreasing unmet demand in this way, it would also have been possible to reduce the amount of goods carried in trolleys by 25% to cut out overstocking with only a minimal impact on sales.
By reducing the weight of each aircraft, the airline could also save on fuel and stocking costs, or could choose to load more items onto trolleys with the prospect of greater sales.