Scan and Supply

U.S. manufacturers have largely rejected demand-driven replenishment-a technique for using actual consumer demand to determine supply quantities-but new Internet capabilities could revive this practice.

By -- Logistics Management, 6/1/2000

For one U.S. jeans maker, demand-driven replenishment has proved to be a good fit. VF Corp., which manufactures Lee and Wrangler jeans, now supplies some 13,000 stores across the United States based on point-of-sale (POS) data. "When we first started the program," says Jeff Kernodle, VF Corp.'s vice president of replenishment, "the normal time [from re-order to delivery] was 30 days. The average today is seven to 10 days."

Although the Greensboro, N.C.-based VF Corp. has reduced the amount of inventory in its pipeline significantly through demand-driven replenishment, it's one of only a handful of U.S. companies using this logistics strategy. Demand-driven replenishment was the "Holy Grail" of supply chain management a decade ago, but the concept has languished in recent years.

Concerns about data integrity and mistrust between merchants and manufacturers have made implementation of demand-driven replenishment a tough sell. "Conceptually, demand-driven replenishment using actual POS data is valid," says James Hintlian, a partner at Andersen Consulting in Wellesley, Mass., "but in practice it has fallen short." Nonetheless, some industry experts think that the push toward electronic payment of manufacturers-called scan-based trading-might revive the moribund strategy.

High Hopes

When the retail industry first broached the concept of Quick Response (QR) and its grocery-industry counterpart, Efficient Consumer Response (ECR), a decade ago, many of the practice's advocates envisioned that data on consumer takeaway eventually would determine product resupply in the stores. The widespread deployment of the Universal Product Code-a bar code stamped on every product-made this concept possible. As clerks scanned items at the checkout counter, the retailer would gather data that it could then share with the manufacturer. The exchange of information on consumer demand would, in turn, guide the manufacturer's production and restocking activities.

One of the pioneers in the grocery industry, Shaw's Supermarkets Inc. of East Bridgewater, Mass., actually began using point-of-sale data to determine daily warehouse withdrawals as soon as it installed scanner systems for reading product bar codes in 1976. "We were one of the first in the industry to do this," says Shaw's spokesman Bernard Rogan. "As soon as the scanning system was put in, Shaw's set up a mechanism to replenish off scans. Now it's part of the culture."

The grocer currently uses point-of-sale data to determine what items to ship each day from its four warehouses to more than 160 stores throughout the Northeast. Although it uses the POS data internally, however, Shaw's does not share the information with manufacturers directly. "That's our information," says Rogan, "and we prefer to keep that to ourselves."

Although Shaw's is reluctant to share point-of-sale data, some retailers have been more open. VF Corp., for example, says it receives data from some store chains. "We've been doing this with Wal-Mart for 10 years," says Kernodle. "Having access to what's selling at the retail level provides us with the information we need to do a better job of forecasting and planning."

Kernodle says that demand-driven replenishment does not mean that "we sell one and ship it back." Rather, he notes, point-of-sale data are used to refine VF's forecast of future demand. "We generate an order based on the store's meeting a minimum need. We've been successful working with a lot of accounts because it cuts time out of the replenishment cycle."

As it turned out, the adoption of demand-driven replenishment brought about major changes in VF Corp.'s distribution operation. For one thing, the company makes more frequent shipments to retailers from its six distribution centers across the country than was formerly the case, often sending items directly to the stores themselves. In the past, it typically shipped large quantities of single stock-keeping units. Today, however, it ships assortments of product sizes and styles to match expected store-level demand. In order to be able to build these assorted shipments, the company has largely automated the pick-and-pack operations at its distribution centers. "Ninety-eight percent of what we pick and pack today is for a given store to satisfy consumer demand," Kernodle reports.

Data Differences

Although a few soft-goods makers are resupplying stores based on actual sales data, most have not gone down that path for a couple of reasons. "Five to seven years ago, the apparel and soft-goods industry tried POS-driven replenishment," says Charles Troyer, a consultant with Cleveland-based CSC. "But companies found it was difficult to maintain inventory at the accuracy level required. If you're off by one or two cases in a warehouse, it doesn't affect you much. In a store, if you're off by one or two cases, you're out of stock."

A lack of detailed sales data also has bedeviled attempts to promote this strategy. Data corruption often occurs at the checkout stand. Take the case where a customer buys 10 jars of baby food, each a different kind. Instead of scanning each jar, the clerk takes one jar-say peas-and scans it 10 times. The store's computer records a sale of 10 jars of peas instead of a mixture of products.

Retailers' distrust of manufacturers has also hindered demand-driven replenishment's adoption. "The retailers are hesitant to give away POS data for free," notes Chap Kistler, a partner in the retail consumer-products practice at the consulting firm Ernst & Young in New York City.

Manufacturers also worry that they would incur extra handling costs if they were to pick and ship products specifically for individual stores. "From a manufacturer's standpoint, converting to demand-driven replenishment means you have to pick store-specific lots," says Kistler. "Most manufacturers are not prepared for this."

In fact, because of these issues, manufacturers and retailers have turned to Collaborative Planning, Forecasting, and Replenishment (CPFR) as an alternative to demand-driven replenishment. Under CPFR, the manufacturer and retailer compare sales forecasts and then arrive at a joint plan for production and deployment of a specific product. "Companies are looking at macro-level planning activities that smooth out the lumpiness in demand, and that's where Collaborative Planning, Forecasting, and Replenishment comes in," says Hintlian of Andersen Consulting. "Trying to micromanage replenishment is difficult."

Scan-Based Trading

Although CPFR has eclipsed demand-driven replenishment for the moment, some industry experts believe that the recent emergence of scan-based trading could revive the concept. Under scan-based trading, a retailer pays the manufacturer promptly based on what's scanned at the checkout counter.

Dreyer's Grand Ice Cream in Oakland, Calif., has engaged in scan-based trading with some of its retail customers for the past six years. Still, the ice-cream maker doesn't use point-of-sale data to drive replenishment; instead, POS information is used to generate forecasts for direct store delivery. "We're actually capturing movement data and then using that to forecast demand in a forward sense," says Michael Corby, director of distribution development at Dreyer's, which makes such premium ice-cream brands as Edy's, Godiva, and Starbucks.

At present, Dreyer's is one of a half-dozen manufacturers taking part in a Grocery Manufacturers' Association-sponsored pilot on full scan-based trading, using an extranet as the conduit for data exchange. The six manufacturers are working with grocers Schnuck Markets of St. Louis and Andronico's Market in Albany, Calif. Results of the pilot, which was started in June 1999, are expected to be released this summer.

The pilot's participants hope to show that scan-based trading can result in item pricing synchronization between the manufacturer and the retailer. In addition, the pilot should help them determine whether a manufacturer and retailer can eliminate backroom check-in, saving time for both parties. Elimination of merchandise check-in by store personnel would presumably increase routing efficiency because manufacturers would be able to deliver shipments during off-peak hours.

The potential advantages notwithstanding, the six manufacturers are still not restocking store shelves on the basis of point-of-sale information. "Right now it's strictly billing," says Tom Nealon, chief information officer for Frito-Lay Inc. in Dallas, which is also taking part in the pilot. "We have not taken it into demand-driven replenishment. There's no point doing that until you know you have the pipeline working."

A Return to Demand-Driven?

Despite the obstacles, some of those participating in the scan-based trading pilot foresee a day when that practice could evolve into actual demand-driven replenishment. "It's theoretically possible," says Corby. "The issue comes down to the number of variables you have to manage. The day-to-day replenishment would be easy. It's the other events-such as promotions and holidays-that have so much variability."

In the long run, however, most still view consumer-driven replenishment as the shining goal for supply chain management. "It's inevitable that POS data will be used to obtain supply chain efficiencies," says Nealon. "It's just not happening today."

How scan-based trading changed distribution at Dreyer's