AI Case Study
DM drugstores optimises shift planing in retail stores using advanced prediction modelling
German based DM drugstores implemented long-term prediction models of sales based on inputs such as historical sales data, opening goods, new product introductions, holidays and even local traffic diversions. Planning shifts to meet customer demand has traditionally been a difficult task for store managers. Either over or under staffing can create issues with unnecessary expenses, poor customer service and staff satisfaction. The models were strong enough to allow staff to book their shifts well in advance with few last minutes changes.
Consumer Goods And Services
Food Beverage And Drugs
"The approach was to implement a long-term prediction of daily store revenues, taking into account a wide range of individual and local parameters. Input data to a new algorithm included historical revenue data, opening hours, and the arrival times of new goods from the distribution centers. On top of this, other data was ingested to achieve the highest level of precision. This data included local circumstances such as market days, holidays in neighboring locations, road diversions, and – in future – weather forecast data (as weather conditions significantly impact consumer behavior). DM evaluated different predictive algorithms, and the selected solution now provides such accurate projections that it has proved to be a powerful support for shift planning."
"Based on the high-resolution prediction of daily sales for each individual store, employees can now enter their personal preferences into the shift schedule four to eight weeks in advance. Once approved, their shifts are unlikely to change; they can rely on the long-term plan, and a last-minute change is an exceptional event.This shows how applying predictive analytics at DM is increasing in-store operational efficiency and, at the same time, is contributing to a better work-life balance for store personnel."
"At DM drugstores, the shift planning task was historically performed by the store manager based on simple extra-polations and personal experience. For regular business days, this process was good enough. But with an increasing number of exceptions, it became unsatisfactory. Over-head or shortfall of personnel limited store performance.So DM determined to effectively assist store managers in their personnel forward planning by finding ways to reliably forecast demand at each particular point of sale."
"DM implemented long-term prediction models of sales based on inputs such as historical sales data, opening goods, new product introductions, holidays and even local traffic diversions."