CAN COMPUTERS FIX THIS?

Julie Gallagher / Additional reporting: Michael Garry

In this age of automation, it is commonplace to allow computers to automate processes previously handled manually by humans. When it comes to replenishing goods on store shelves, however, food retailers have been very careful about making this transition.

Some retailers continue to use "eyeballing" of holes on store shelves as the basis for placing replenishment orders on handheld terminals, padding the order for promotions. Yet with out-of-stock conditions continuing to bedevil retailers, more are beginning to explore a computerized approach, which allows them to anticipate and prevent less-than-optimal stock quantities.

Two types of computer-based ordering scenarios have emerged: one conservative and guided by manual intervention (computer-assisted ordering, CAO); and the other willing to rely primarily on technology to make ordering decisions (computer-generated ordering, CGO).

Both scenarios employ forecasts of consumer demand. CAO uses it to produce a list of order suggestions, while CGO generally combines the forecast with data on a store's current stock, or perpetual inventory, to automatically push orders through to the warehouse or manufacturer.

To date, retailers have been more successful in automating replenishment of their distribution centers than their stores. "At the distribution center level, we've gotten very, very good at replenishment, but at the store level, retailers continue to struggle," said Barbara Anderson, president, BVAC, San Carlos, Calif., and an author and expert on computer-based replenishment.

"As an industry, we understand how to handle factors like promotions and seasonality at the DC, but when we go from the same method to the stores, it's times 1,000."

Also, store automation requires retailers to exercise tight store discipline over areas like scanning and receiving.

Still, some technology-oriented retailers are succeeding with computer-based replenishment, helped by the growing availability of third-party software aimed at these processes.

Price Chopper Supermarkets, Schenectady, N.Y., has worked on both CAO and CGO over the past few years, employing an application from IMI Americas, Mt. Laurel, N.Y. Ukrop's Super Markets, Richmond, Va., has partnered with DCM Solutions, Irving, Texas, in the development of a demand-based replenishment solution.

For the past few years, Hannaford Bros., Scarborough, Maine, has been pursuing what Bill Homa, its chief information officer, called "a poor man's CAO" for the ordering of perishable goods.

The internally developed system allows headquarters to tell stores "what the sales for tomorrow's delivery were last week and last year," and suggest order quantities, said Homa. "It helps build a better order at the store for fresh." Nonperishable orders are handled in the traditional "eyeball" manner, though headquarters lets stores know if a product is out of stock at the DC.

Hannaford plans to replace its home-grown CAO with a "real CAO" system, from SAF, Tagerwilen, Switzerland, which would apply to both perishables and nonperishables, Homa said. The chain is currently piloting the SAF system, which requires considerable information from various systems to be "readily available," he said.

Influenced by the automation of its warehouse ordering processes, Colemans Food Centre, Corner Brook, Newfoundland, moved from a forecast-free, manual replenishment system to a computer-assisted process for in-store ordering of its groceries and frozen foods.

Colemans' model is supported by INGEN, an application marketed by Distributed Intelligence Systems, Lake Mary, Fla.

"Our strategy is to move as close as possible to a just-in-time inventory, while optimizing layouts and categories within the stores," said Scott Bennett, director of information technology, Colemans.

Although it has greatly reduced store inventory and out-of-stocks by moving from the labor-intensive process, the 11-store retailer probably won't ever completely automate its ordering functions, said Bennett.

"No matter the strategy or parameters put into the system, there are always going to be times when the store manager will have to override or adjust [an order]," he said. "There is too much risk not to review orders."

Colemans is not alone in its cautious approach to store replenishment automation.

Although a large percentage of grocers have taken steps toward automating ordering processes, according to Anderson, a low percentage is successfully running truly automated ordering processes.

Anderson suggested that an automated replenishment strategy may be more desirable than a computer-assisted model because it frees up store personnel and compensates for the inefficiencies employee turnover can create. Automated systems may even produce a more optimized order, she noted.

At Colemans, according to Bennett, last-minute order changes are sometimes necessary due to unforeseen circumstances, including competition, weather, customer requests for product, and new products, which lack sufficient historical sales information. Colemans reviews each of the order suggestions provided by the system.

The order suggestions are based in part on the store's perpetual inventory, which the retailer began monitoring when it implemented the computer-assisted replenishment system.

Sales forecasts, which are created for each stockkeeping unit, are based on pricing and promotional needs, historical information, time of year, shelf space, and the desired minimum amount of stock for a particular item, explained Bennett.

Since moving from the manual processes where orders were based solely on product holes and back-room inventory, Colemans has a greater ability to focus on customer service.

"The labor component for the manual ordering was quite intense," said Bennett, "whereas now labor is still required to look over each order, but it's a lot shorter process. Our store managers tell us they'll never go back to the old [ordering] method."

Going All the Way

One retailer that has graduated from a computer-assisted ordering system, similar to the one used by Colemans, to an automated CGO system is South Africa-based Massdiscounters, which wanted to cut down on the labor dedicated to the previous system and ensure more accurate orders.

Massdiscounters, a 67-store discount retailer, operates under the Game and Dion brands, and offers a range of general merchandise, as well as nonperishable groceries.

Before implementing Irvine, Calif.-based Just Enough Software's Forecast-to-Replenishment solution, Massdiscounters' order suggestions were produced by a home-grown system. Under the earlier system, orders were created overnight so they could be reviewed and processed by the retailer's personnel the next day, according to Mervyn Naidoo, replenishment executive, Massdiscounters.

"Because of our diverse range of products, there was an estimated 40% intervention rate" under the former system, he explained. "The demand was erratic, and the system wasn't recommending suitably accurate orders. If you're continuously intervening, you're never going to get the stock turn that you desire."

Massdiscounters' current system combines historic and current price and promotion information to automate the functions of forecasting, inventory optimization and replenishment.

As a result, order intervention is only necessary about 2% to 5% of the time, and Massdiscounters has achieved an inventory accuracy rate of between 90% and 95%, according to Naidoo. He described the savings associated with the system's efficiencies as "enormous."