Optimize Your Ordering Process

An out-of-stock is more than just an empty shelf. In the consumer's mind, an out-of-stock occurs any time they come into the store ready to buy, but leave without purchasing an item for a reason other than it is priced less elsewhere. Some retailers rely on their IT systems to tell them what their out-of-stock levels are. While this might be a satisfactory solution for the enterprise as a whole, it does not take into account the realities faced by retailers and consumers at the store level. Sometimes, corporate-level ordering decisions do not meet the varying demands of each individual store. Mike Holcomb, director of IT services at Spartan Stores, dealt with this exact problem. Not only did he have to automate forecasting and replenishment, he also needed to house that solution at the store level rather than at corporate.

Based in Grand Rapids, MI, Spartan Stores owns and operates 100 supermarkets in Michigan and Ohio and employs 10,000 people. The grocer operates under the D&W Fresh Markets, Family Fare Supermarkets, Felpausch Food Centers, Glen's Markets, VG's Food, and pharmacy banners. Spartan Stores is not a big-box grocer — some stores are 17,000 square feet. The stores did not have the typical systems that many big-box retailers operate, such as CAO and perpetual inventory (i.e. an accounting method of maintaining up-to-date records that accurately reflect the level of goods on hand). Spartan Stores replenished products manually, which caused a host of problems relating to inventory, out-of- stocks, and product variety.

Take The Guesswork Out Of Replenishment
Spartan Stores completed the order fulfillment process two to three times per week based on product volume. Each store is separated into four nonperishable departments: frozen foods, dairy, general merchandise, and grocery. There are four order writers (i.e. employees in charge of order replenishment) — one for each department in every store. Therefore, throughout 100 stores, the grocer employs 400 order writers. Spartan Stores' order writers placed orders for low or out-of-stock items by walking up and down the aisles with a Honeywell handheld scanner. "The order writers made the ordering decision based on skill and years of experience," says Holcomb. "For instance, an experienced order writer [someone with more than two years of order writing experience] might place an order based on their knowledge that a shelf holds 10 units of a product." The process was inefficient because experience and skill level fluctuated among order writers. Consider an instance when an order writer could not come to work on order fulfillment day. Someone else with less experience or their own method of fulfilling orders would complete the task, which created grave inconsistencies in the replenishment process. It was also difficult to determine back room inventory before placing orders. The stores may have had inventory in the back room, but order writers had no way of knowing, as an accurate, up-to-date perpetual inventory system didn't exist. Spartan Stores used sales data reports generated from POS systems to help guide orders (e.g. 120 units were sold the last time this item was on sale).

Ordering problems occurred for other reasons as well. Holcomb recounts instances when a customer asked an order writer for assistance, and the order writer left the aisle to assist the customer. When he returned, he'd forgotten where he left off and therefore skipped entire sections of product. The items he skipped ran the risk of being out of stock because they weren't reordered until the next order process. In addition, shelf tags, which contain bar codes that order writers scan for reorder, often fell off shelves. If a shelf tag was missing, order writers didn't order that item, which created out-of-stocks and a lack of variety. For example, order writers might not replenish Cinnamon Life cereal because its shelf tag is missing, leaving only regular Life cereal for purchase.

Implement A Forecasting Application At The Store Level
The inconsistencies brought on by manual ordering led Holcomb to search for an automated CAO and perpetual inventory software system. Based on the grocer's business model of different markets (i.e. rural, urban, seasonal), Holcomb wanted a system that used only store-specific information to generate orders for each store. He felt each store should place its own orders and send that information to corporate, instead of allowing corporate to make ordering decisions. "Our stores have unique product needs," says Holcomb. "Even a sophisticated inventory system would not provide the individual store-level visibility we needed. We needed a forecasting application in every store."

Holcomb purchased a CT2020 software application from SofTechnics. CT2020 is a suite of wireless, mobile, in-store applications using shared data to enable retailers to gain control of their store-level inventory, product replenishment, and item and price management-related activities (see sidebar on page 20). He chose SofTechnics as the grocer's perpetual inventory vendor because the vendor already supplied the grocer's item management, direct store delivery system, and shelf verification modules. However, Holcomb did a formal RFP for a CAO application with two companies. The grocer chose SAF SuperStore, because the application can be implemented at each individual store. The application uses only store-specific movement information to generate orders for each store. Also, SofTechnics partners with SAF and integrates CT2020 and SuperStore prior to the store-level implementation. CT2020 keeps a running inventory for every SKU within the four walls of each store. It does so by tracking all receipts, sales, and adjustments to every SKU. Based on all movement history and other demand- influencing factors (e.g. removes negative inventory since a store can't have a negative amount of products), the automated forecasting application uses algorithms to calculate a forecast to determine product for a demand period. SuperStore presents the forecast information to the order optimization module. The order optimization module places the order based on demand. Finally, CT2020 sends the order information to corporate via a WAN.

An SMB retailer automates inventory at ismretail.com/jp/7524.

Work Out Kinks With Phased Implementation
Spartan Stores took a phased approach to the perpetual inventory, CAO, and forecasting applications implementation. Beginning in September 2006, Spartan Stores conducted a two-store pilot. The frozen foods department went live first, followed by the dairy department. In April 2007, the grocer brought on the grocery department, followed by general merchandise in August 2007. By February 2008, the grocer had rolled out two additional stores. At that point, Holcomb measured results for the pilot stores to convince Spartan Stores' executive team to approve enterprisewide rollout. The executive team approved the entire project, and enterprisewide rollout began in August 2008.

Spartan Stores' IT team performed all application installations at each individual store. The IT team installed the software on each store's IBM servers and configured the application for each store. Each installation took 3 to 4 hours. Spartan Stores' Retail Operations team provided training for everyone who uses the product. The store operations team and department managers were trained first, followed by cashiers. Training was critical, as CT2020 and SuperStore have completely altered the way products are ordered, returned, etc. For example, if a customer returns a product, it must be returned at the UPC level, not at the department level, as it was previously.

Since the applications are a major cultural change to the organization, Holcomb encouraged each store operations team to participate in the process. "This project creates an immense cultural shift, because you're having a computer take over for an employee who's done the job for 15 years," says Holcomb. "Yet, the store operations teams were the best proponents of the application because they knew these systems would drastically optimize the ordering process. We did not intend to reduce labor as a result of this project." Spartan Stores reassigned order writer labor to other pressing store functions, like overseeing perpetual inventory.

As of press time, Spartan Stores has implemented the SofTechnics CT2020 and SAF SuperStore in 20 stores. To date, the retailer has reduced out-of-stocks by 65% and has reduced inventory nearly 10%.  Holcomb states employees are satisfied with the system, because they no longer spend time on the tedious manual ordering process.

Implementing a store-level CAO system will help every store meet its own unique demands. As a result, customer satisfaction and loyalty will increase because Spartan Stores offers the products customers want when they want them. In addition, warehousing costs will decrease because order numbers are accurate.