Potential Improvements of Sales Forecating Practice in Retail: the Case of Wilko Retail LTD

Sales forecasting is vital for any company, especially in retail. Although the importance of promotional sales forecasting is recognized by all retailers, only some of them utilize professional software, sticking to simpler ad-hoc methods. Even specialized forecasting softwares are often implemented by external contractors who disregard the specifics of the company and its business, which results in a poorly performing forecasting model. The companies though prefer not to change the model, but to use extensive manual corrections, taking extra time and creating a new source of errors. This study focuses on the case of Wilko Retail Ltd, one of the UK largest household and homeware goods provider, and namely the improvement of its forecasting procedure. In agreement with the company three objectives were specified: • Development of a new forecasting performance procedure; • Evaluation of the currently applied model and identification of improvement opportunities; • Development of solutions to the identified bottlenecks and their integration into the model. The outcomes of the project include: • A new forecasting performance assessment procedure measuring not only accuracy, but also bias of the model; • A model, outperforming the their system by 3.22%, while requiring less historical data, not involving any manual adjustments and providing the best results at the periods important for any retailer: Christmas, New Year and Easter. Such innovation is vital for Wilko facing problems both in the market (growth of competitors, increasing turnover bringing no additional profits) and macroeconomic situation (Brexit, introduction of a national wage, depreciation of the sterling) because it would become a source of considerable cost savings due to the improvement of the whole distribution and replenishment systems. These cost savings will contribute to its cost cutting initiative launched in 2017, while still saving Wilko employees’ jobs.