INTEGRATING DEMAND FORECASTING WITH REPLENISHMENT IN A HIGH-TECH RETAIL CHAIN

Headnote

Since many of the customers of the high-tech industry are witting to wait when retailers are out-of-stock, stores should keep inventory at a low level ... often shortages occur because of the poor replenishment procedure at the central warehouse ... the more volatile the sale of a product, the more difficult it is to forecast.

Demand forecasting and efficient replenishment policy are two important elements in managing inventory, a difficult challenge in the high-tech industry. High-tech products are characterized by short life cycles, too many SKUs, volatile demand, and quick obsolescence. As such, demand forecasts are unlikely to be accurate, resulting in product shortages and lost sales. To counter such issues, high-tech companies carry extra inventory, which is very costly.

In this article, we propose an experimental approach to address demand forecasting and efficient replenishment in a high-tech retail chain. The approach consists of three steps: (1) selection and evaluation of the demand-forecasting model, (2) integration of the demand-forecasting model with the chain's replenishment policy, and (3) application of these two models to a pilot program. The experimental approach has been applied to the Greek Telecommunications Organization, which operates a wide retail chain of stores that sell high-tech telecommunication and mobile telephony products.

THE REPLENISHMENT PROCESS

The high-tech retail chain comprises a number of stores across the country and one central warehouse. The stores sell high-tech telecommunication products, such as telephones, phone cards, and ISDN equipment. The replenishment process of the chain is periodic (see Figure 1); on a given day of every week, store managers place their orders to the Central Office (COF). The COF checks the orders and forwards them to the Central Warehouse (CWH). Once a week, the CWH sends batch orders to the suppliers, e.g., Nokia and Ericsson. The CWH delivers the products to stores based on their orders and availability. The ordering process does not utilize any forecasting technique.

The retail chain faces two main problems: low product availability (Out-of-Stock), which leads to lost sales and decreasing customer loyalty; and increased safety stock, which is accompanied by high holding costs and increased risk to sale stock at depreciated values.

THE EXPERIMENTAL APPROACH

We have developed a five-phase experimental approach for integrating the demand-forecasting model with the high-tech chain's...