Consumer loyalty is the key to profitability in today's fiercely competitive, multi-channel retail market, and earning that loyalty depends on having the right products available when and where shoppers want to buy them. As noted in the Annual Consumer Outlook report from Kurt Salmon Associates, "Consumers cannot be loyal purchasers if they can't trust that retailers will consistently have the merchandise they are looking for."

A mandate to have the three "Rs"- the right product in the right place at the right time - certainly is nothing new for retailers, but meeting that goal continues to be a challenge.

"The vast majority of retailers today are still ordering from the store by walking the aisles with their hand-helds and looking for products low on stock," says William Alexander, IBM's worldwide manager of supply-chain solutions for the retail industry. This system not only is inefficient, he says, but also is highly susceptible to human error. And those retailers that have started doing computer-assisted re-ordering are not a lot better off, since they typically mix scan-based data with historical information to create a forecast. "Getting a good forecast at the individual store level is really, really difficult," he says. "There are just too many items and too much happening on a daily basis."

Consequently, most retailers experience a significant number of lost sales due to product stock-outs. At a recent grocery industry conference, three out of four retailers surveyed put the figure at more than 5 percent of total sales. "It is a fact that retail stores, despite more and more technology, still experience out-of-stocks anywhere from 10 percent to 12 percent across their product range," says Alexander.

That's because technology solutions to date largely have focused on product availability at the distribution center rather than the store, he explains. Many consumer goods companies now report over 99 percent availability at the DC, but have been unable to achieve similar results at the highly unpredictable store level. Solving replenishment across these "last 100 yards," he says, "is where retailers have the greatest opportunity to improve customer service and drive store revenue."

To address this area, IBM and Industri Matematik International, a Sweden-based software company with U.S. headquarters in Mt. Laurel, N.J., jointly developed a new software product called e-Replenishment. Designed to run in high-volume, highly automated environments like grocery or drugstore chains, e-Replenishment takes scan-based point-of-sale (POS) data and uses it to immediately replenish what actually moves off the shelf. "It is a just-in-time solution for retail," says Alexander.

Electronic POS data long has been used as a factor in developing demand forecasts, but few attempts have been made to have the data directly drive store replenishment. One reason is that data accuracy is compromised unless good procedures are designed and consistently followed. At the checkout counter, for example, clerks have to be trained to scan every item, every time: 10 cups of yogurt may be the same price but represent a combination of flavors and SKUs. Item master files also must be rigorously maintained.

Through various methods, including in some cases disabling a clerk's ability to hit a multiple-item button on the register, most "forward-thinking, fast-moving retailers" have mastered this problem, says Rob Sweeney, director of product strategy at IMI. For them, "data accuracy is no longer an issue."

Other barriers - the very high volumes of transaction data associated with retail and the technical infrastructure needed to effectively transport this data from store systems in real-time - have been solved by the internet and advances in messaging architecture.

The e-Replenishment product grew out of separate work IBM and IMI were doing with a joint client: Albert Heijn, the largest grocery chain in the Netherlands, and its parent Royal Ahold, one of the world's biggest retailers. When the two software companies formed a strategic alliance last June and began exploring areas where they could work together, their efforts at Royal Ahold provided a natural field for collaboration.

Albert Heijn was working to implement a program that it calls "Today for Tomorrow," in which the goal is to use today's POS data to replenish stores the next day. IMI had been brought in to provide necessary fulfillment center and warehouse management operations and IBM was working with Ahold on this and other projects. "As we began to understand what the customer was trying to do, we realized at IMI that we had an applicable offering much broader than just solving problems in the warehouse," says Sweeney. The company subsequently sold Albert Heijn its order management engine. And like any software company that helps develop an innovative solution, "we began to think about expanding this application to a larger market," says Sweeney.

E-Replenishment uses Advanced Order Management and Fulfillment Center software from IMI's Vivaldi suite of products for execution of the replenishment process. IBM's e-business framework, in particular its MQ Series family of middleware, provides the high-volume messaging backbone that connects Vivaldi with retail stores and the extended supply chain.

Implementation begins with an initialization process, which assesses the shelf capacity required for each item to ensure that availability to the shopper is maintained between replenishments. This assessment takes into consideration such factors as historical consumption patterns, existing replenishment cycle times, pack sizes for distribution, and display and space-management considerations. "Most of this information we would expect a retailer to feed into the program from other applications, such as a merchandise planning system," says Sweeney, though it can be done manually. Goals are set for key parameters that will drive the replenishment process: replenishment cycle time, and economic replenishment quantity, such as a case size, and service level.

The next stage is actual execution, which starts with the input of scanned data from purchased items into a POS system. The Demand Accumulator, a component of the Vivaldi order management software, amasses this data in a real-time continuous flow. Using the parameters set during the initialization process, Demand Accumulator recognizes when product should be replenished and sends a message to Advanced Order Management, which automatically creates the replenishment order or adds to an existing order. If needed, the Vivaldi system also performs other rules-based tasks, such as determining how each line item on a replenishment order will be sourced. Once a replenishment order is created, Vivaldi provides visibility to all parties involved.

Vivaldi subsequently releases replenishment orders to the points of supply for fulfillment and shipping, usually for next day delivery. "The only gating factor is the efficient quantity to ship, which is probably a case," says Sweeney. Depending on this factor and cycle time, some stocks may be replenished "day after tomorrow," while it is conceivable that very fast moving items would be replenished the same day.

While e-Replenishment operates automatically on the vast majority of items, the process is monitored. A simple browser-based graphical interface is used to manage exceptions that need human intervention. Unexpected spikes in demand, for example, might necessitate a change in the trigger point for replenishment. IBM's business intelligence technology and a browser interface supported by its WebSphere platform allow retailers to quickly diagnose problem situations and take appropriate action.

The monitoring activity also enables continuous improvement, say Sweeney and Alexander. Tracking the exceptions makes the benefits of cycle-time reduction and pack-size reduction for certain items more apparent and easier to sell within the extended enterprise.

Initial results from this type of direct replenishment system are impressive. Albert Heijn still is in the implementation phase for its 680 stores, but a similar "availability-based replenishment" system was installed in just six months at 16 supermarkets and six hypermarkets of Guyenne and Gascogne, a franchisee of French retailer Carrefour. After implementation, product availability went from 88 percent to 98.5 percent. "Really the only reason they are out of stock now is if the DC or supplier is out of stock," says Alexander. Sales per square meter also increased significantly, as much as 30 percent in some categories, the company reports. Moreover, even though there was no specific focus on inventory reduction, store inventory dropped by 21 percent. The number of employees involved in purchasing was cut by 80 percent, from four people to one. These results helped make Guyenne and Gascogne Carrefour's best performing division last year.

For a large retailer with 500 stores that average $28m each in gross annual sales, IBM and IMI project that e-Replenishment could achieve a gross profit increase of $280,000 per store per year or $140m across the chain. Inventory investment, they say, could be reduced by $160,000 per store or $80m across the chain. (See chart.) This is based on a relatively conservative assumption of a 5 percent availability improvement, which would drive a 5 percent improvement in sales.

"You don't have to work through the numbers very much to see the immense value potential here," says Alexander. "This is particularly true if you think about how difficult it is to get growth in these sectors in the U.S. and Europe at the moment because they are such mature markets."

The solution, says Sweeney, also would be applicable to large home-improvement and office-supply stores as well as to many specialty stores.

Demand Forecasting
For other retailers, an advanced demand forecasting solution still is the best way to improve in-stock positions. Saks Inc., which operates 360 department stores under various names in addition to Saks Fifth Avenue, is implementing Logility's Voyager software across its entire system to improve replenishment of basic stock items like socks, underwear, jewelry, home-store items, and many other products.

The impetus behind this move "is to make sure we are maximizing the sale of basic items, which are very high gross," says Marty Abercrombie, vice president of replenishment for Saks Inc. "These items are not nearly as susceptible to markdown as fashion items, and it is important that we don't lose sales because we are out of stock." Customer service issues also are involved. "We want to maintain a reputation for high service so that customers know they can come in a store and find the basic items they are looking for. Then, while they are in the store, hopefully they will also take advantage of our fashion offerings."

When Abercrombie came to Saks in the fall of 1998 to manage basic replenishment, there was no systematic process in place to set model stock levels for the 7 million to 8 million store stock locations under his charge. Inventory targets for these locations were set using historical sales figures and assumptions about selling rates. After benchmarking industry best practices, "we laid out what we wanted a replenishment system to do," says Abercrombie. "We found that the key pieces we were missing were a sophisticated sales forecasting tool and also a dynamic modeling tool."

The latter, Abercrombie explains, is a system "that would automatically adjust our model stock levels for different items at different stores without us having to manually go in there and reset the levels." Such a system would have the "intelligence" to recognize seasonal patterns as well as changing consumer-buying patterns.

Saks also wanted a system that could easily accept and process information about changes in promotional strategy, and that was where Logility really stood out, says Abercrombie. "We wanted a system that would offer us the ability to influence the forecast at a higher level and then have that change pro-rated down to all different store SKU locations that would be impacted."

In contrast to many systems that focus on forecasting at the store level only, says Abercrombie, Logility has a pyramid forecasting strategy that lets users forecast demand for product at the store and corporate level, and by style, group and department. "The reason this is important is because our demand can be sporadic or very low-volume at the store SKU level. A given item may only sell one time in four weeks and it is hard to pick up patterns at that level. But when you accumulate demand and forecast it at a higher level, it starts to get more predictable. You start to see seasonal patterns and trends."

Once Saks develops a good forecast with Demand Planner, it will feed that information into the Logility inventory planning system. "We will put inventory rules into that system for each item - how many weeks of supply we want, what kind of service level we want to achieve, what is the lead time for the vendor. The Logility inventory planning systems then takes all that into account to determine what the target inventory level or inventory model should be for a particular store."

Saks anticipates a 1 percent increase in sales system-wide as a result of better basic-item forecasting with the Logility system, says Abercrombie. The company expects to have the system installed and running at all its department store operating companies by the end of this year and at Saks Fifth Avenue by January of next year.

The next step will be sharing and fine-tuning its forecast with suppliers, using Logility's collaborative capabilities. "First we want to get the basics down. We did not have a forecasting system before we purchased Logility and we want to get a good forecast, one we are comfortable replenishing to, then we will take it the next step," Abercrombie says.

He emphasizes, however, that sharing information with vendors is an important part of Saks's longer-term strategy. "Once we are forecasting effectively and have good inventory policies for our items that drive good inventory targets for each store location, it does us no good if the vendor can't ship product to us when we need it. That has been an issue we have faced in many product categories, so we want to do everything we can to help vendors know what our needs are. It's all part of our goal of maximizing sales by being in stock on basic items."

Logility recently introduced two new tools to facilitate collaboration between trading partners over the internet. Voyager XPS, extensible planning solution, is an open architecture collaborative tool that is compliant with industry standards for Collaborative Planning, Forecasting and Replenishment (CPFR). XPS supports Logility's i-Community marketplaces for vertical industries. The marketplace for the furniture industry already is operating, with Wickes furniture, a $350m retailer with 30 stores being the first to sign on.

Another new product, Voyager XES, or extensible execution systems, brings the carrier into the collaborative process. "This is a natural extension of CPFR," says Andrew White, vice president of Logility. "It allows carriers to gain visibility into order forecasts a retailer and manufacturer are collaborating on. They can drop that data down, and allocate capacity, assets and vehicles to the load. Planning assets more efficiently can reduce their costs, and the implicit assumption is that they will share those benefits with their collaborative trading partners."

Demand Convergence
Software vendor Manugistics is focusing on the problem of maintaining in-stock positions across multiple channels, in its latest retail solution, introduced at the International Mass Retail Association's Logistics 2000 Conference in February. "What we are seeing is that consumer-facing companies need to provide the consumer with common experiences, regardless of which channel the consumer chooses to shop in, says Chris V, Verheuvel, manager of strategic solutions.

"The challenge is offering up the same ultimate consumer experience for the shopper, whether in the store, on the web site or via a catalog, and integrating all the back-end logistics pieces to make that consistent branded experience happen across all channels," adds Dawn Andre, vice president of solutions and alliances.

Manugistics addresses this sales channel convergence by linking customer orders and future demand across all channels with the same real-time, available-to-promise information, using the advanced WebConnect features of its WebWORKS architecture. This means that "you know what you have available within all your facilities, you know all your alternate sourcing channels, you know what products you have inbound to your customer-facing facilities, and you know where they are within the transit world," says Verheuvel. "Then, when taking orders in multiple spaces, retailers can coordinate and control how they commit those orders."

When a retailer just had store fronts, the ability to commit was not that important a component of a solution set, he says, "but as consumers start going to multiple spaces and the whole speed of business increases, making commit decisions becomes much more complex and much more valuable to the organization."

Manugistics solutions are used by three of the top four Fortune 500 retailers. Leading retail clients include Canadian Tire, Dayton Hudson, Helig-Meyer, Lowes, OfficeMax, Payless Cashways, Ross Stores, Staples, Starbucks, The Gap and Limited.