ICA is the largest general dealer in the Swedish market as well as the largest supplier to the independent ICA-dealers. The principal and one of the biggest stakeholders of this thesis is the department Supply Chain Development who is responsible for the development of ICA’s Supply Chain.

The purpose of this study is to examine how ICA’s sales are affected by weather and to make recommendations on how ICA should proceed with its work regarding this issue. By leveraging their forecasts by including weather, ICA hopes that their logistical metrics amount of spoilage, service levels and number of inventory days will improve. This thesis describes how ICA is working with weather currently, and furthermore it discusses the existing literature within this area. Extensive statistical surveys have been conducted in order to identify the products that are affected by weather, and to examine how they are affected by different weather variables. Finally, the thesis gives recommendations to ICA on how they can improve their forecasting by taking this study’s results into account.

Presently ICA has no consistent approach when it comes to considering weather’s impact on sales. Each individual product planner uses his own experience and gut feeling and there is no central guidance or other quantitative information there to help them.

This thesis is organised around two major statistical studies. First, a correlation study examines the correlation between sales and deviation from mean temperature. The weather parameter was selected following recommendations from earlier studies and from SMHI - the national weather institute in Sweden. The study was conducted on ICA’s entire assortment and identified which items that were affected by weather. These turned out to be about a hundred in number and were affected mainly during the summer; however a few items were also affected in the winter. Most of the affected items were rather logical regarding weather’s effects on sales and have also been pointed out by planners as products that are likely to be affected by weather. Following that, several regression analyses were conducted on the selected weather affected assortment.

The regression analyses were conducted with various combinations of one or more weather parameters as explanatory variables. Surprisingly, in most cases the deviation from mean temperature alone was the parameter that best explained the variance in sales figures. The result or output of this thesis is a value for the expected change in sales for a selected assortment, given a change in deviation from mean temperature.

The recommendations given at the end of this thesis explains how ICA in a simple and cost effective way could implement the study's results in their daily work. The proposal is to divide the work by what should be done centrally by the forecasting department, and by what should be done individually by each planner. The forecasting department should regularly collect weather forecasts over different time periods, calculate the deviation from mean temperature and distribute it to the supply planner. After that, it is up to the supply planners on how to best take advantage of this information by taking into account other unique factors affecting each category, such as campaigns, holidays and daily indexes that in many cases have more effect on sales than the weather.