In just a few years, Internet retailing has emerged as the growth engine of the retail sector. Consider the following.

From 2001 to 2002, Internet retail sales increased 48 percent, defying the U.S. economic downturn. Internet sales were up 30 percent in the fourth quarter of 2003 over the same period in 2002, and sales are projected to increase at a compound annual growth rate of 17 percent through 2008.

Recent surveys find that online purchases account for 32 percent of dollars spent on computer hardware and software, 17 percent of event ticket sales, and 12 percent of expenditures on books. Today, almost two-thirds of all shoppers have used both offline and online channels.

Furthermore, Internet retailing now contributes significantly to profits. Amazon.com reported its first full-year net profit in 2003—some $35 million on $5.3 billion in net sales. That marked a $184 million improvement on 2002 results and was generated from $271 million in operating income. Nearly three-quarters of all Internet retailers have been profitable in recent years.

The opportunities have just begun. Even with the increases described, Internet retailing will account for only 5 percent of total retail sales by 2005. So there's still a long way to grow.

Logistics professionals will be major drivers of that growth. We believe that beyond all the fancy Web sites and 24-hour, in-your-pajamas shopping convenience, it is excellence in distribution, transportation, and inventory management that leads to Internet retailing success.

It could be argued that logistics has been the key to retail success since the earliest days of the Sears Roebuck catalog. Certainly, catalog merchants share a core logistics characteristic with online retailers. When a customer purchases something in a store, usually she takes the product home. It's self-service logistics from the point of sale onward. But an Internet purchase is like a catalog purchase, the e-retailer ships the product to the customer so the responsibility and expense for delivery fall on the shoulders of the retailer and its logistics service providers. (Exhibit 1 summarizes the logistics perspective of physical vs. Internet retailing.)

But unlike catalog merchants, e-retailers can offer more seamless tracking capabilities from order entry all the way to final delivery. Also, Internet logistics managers have to support virtually unlimited product selection, and they must supply perfect information at Internet speed, consistent with the instant feedback that customers expect from Web sites. The Internet implicitly promises to lift all the limitations of catalog shopping, whether in convenience, selection, information, or speed. The logistics manager is on the hook to make this happen.

More and more, logistics has been recognized as a core competence for retailers—particularly with the rise of Wal-Mart and its highly efficient distribution network. But Internet retailing makes the store invisible to the customer. The service encounter with the consumer is now the province of the delivery driver hired by logistics management—not the cashier at the checkout counter. Store personnel no longer stock and maintain the shelves; instead, the logistics manager has to service a Web site's virtual shelves, ensuring that inventory and product information is accurate and up-to-date in real time. Physical store location is no longer critical to retail success since consumers can access the retailer from any wired outlet or through any wireless connection.

But even though stores have been eliminated, physical products still have to be stocked, packaged, and shipped. With store-related costs gone, logistics expenses have become an even larger part of the retailer's cost structure. In 2003, Amazon.com spent $477 million on fulfillment—representing about half of its total operating expenses.

To better understand the influence of logistics in e-retailing, the authors designed and managed a detailed experiment that used "phantom shopper" techniques for purchasing and tracking purchases of music CDs. (This approach is described in the sidebar on methodology on page 60.) Throughout this article, we will refer to findings from that experiment that highlight the central role of logistics in e-retailing, often in ways not considered part of the traditional logistics domain. For one thing, our research confirms that in many e-commerce operations, money is left on the table largely because linkages between customer data and logistics activities are incomplete.

Two things are clear to us. One is that tomorrow's leaders in Internet retail will fully exploit their latent logistics expertise. The second is that logistics professionals must not only recognize their own potential but also become much more active up and down the supply chain. If organizations don't embrace their logisticians' initiatives, their competitors will. (See the sidebar on page 61 on how logistics professionals can support e-commerce).

The research turned up three key areas in which logistics managers can add tremendous value: network design, search and inventory management, and physical delivery. Let's examine each in turn.

Logistics managers have always played a significant role in the design of distribution networks, especially the location of distribution centers to serve clusters of stores. They evaluated the costs and service capabilities of alternative configurations, often using analytical models to get the general locations right. Then they relied on their experience and local knowledge to fill in the details of the precise site, size, and mission of distribution warehouses in the network.

For Internet retailers, logistics realities drive the choice between "pure play" and "brick-and-click" operating networks. Pure-play companies have no physical stores; all deliveries come from warehouses, distributors, or manufacturers. (Amazon.com is the quintessential pure-play, even though some shipments come from other retailers). Brick-and-click retailers display goods and take orders online, offering consumers the convenience of a single point of contact as well as shipment tracking capability. They also have physical store locations, networks, and established logistics expertise that can be employed to satisfy customer requirements.

In our study, we purchased from both categories of retailers. We chose the standard fulfillment option for all purchases so that we received each site's "typical" service. We accounted for individual product and retailer price differences by computing the mark-up percentage for each CD purchased, using data obtained from a trusted CD fulfillment operation. We found a significant difference in the mark-ups obtained by the two categories—mark-ups that underscore logistics' importance.

We discovered that brick-and-clicks have significantly higher mark-ups on a typical CD sale, as seen in Exhibit 2. (Amazon.com and CDnow.com are unusual. The mark-ups of each have benefited from the "links in" from their many affiliates. Additionally, a former Wal-Mart logistics executive helped design and launch Amazon's warehouse network. Amazon has been fanatical in using technology and innovative work methods to contain warehouse and transportation costs. And it has been so successful at using free shipping to build volume that fulfillment costs went from 10 percent of revenue in 2002 to 9.1 percent in 2003 as the free shipping offer was being expanded.)

The pure plays' lagging profitability is not limited to our CD retailers. Webvan and Homeruns.com were spectacular failures in direct-to-home food retailing. Both tried to deliver from dedicated warehouses, and they failed because they could not attain the density of customers necessary to support their local delivery infrastructure. By contrast, brick-and-clicks generally have been more profitable in servicing online customers by using inventories from their conventional stores. These retailers include Tesco in the United Kingdom, Peapod (serving Stop & Shop and Giant customers) on the U.S. East Coast as well as the Chicago area, and Albertsons on the West Coast.

Why the difference? There are multiple explanations, but we believe that many pure-plays have yet to fully understand that logistics expertise is critical to the network design of Internet retail operations. However, at typical brick-and-clicks, logistics professionals have appreciated from the get-go that the "last mile" is a financial disaster waiting to happen without the customer density of a UPS or U.S. Postal Service. They understood that the gross margin necessary to cover delivery costs necessitates extremely large purchases or high prices, either of which limits the Internet retailer's ability to grow.

On the other hand, if a network of stores is already in place, and especially if catalog operations have facilitated learning about home delivery, the direct-to-customer operation has time to grow without having to bear the burden of dedicated people and facilities. In general, our research findings support the idea that brick-and-click logistics professionals have been able to bring their considerable experience to bear to manage the whole delivery activity. They have grasped the complete cost picture. And they have had the relationships with other departments—such as marketing and customer service—that have helped their companies introduce new efficiencies.

Several examples come to mind. Consider how REI's logistics managers have creatively leveraged the store network. Managers at the outdoor-equipment retailer understood from the outset that the Internet was just another channel for gathering orders—a channel that offered new convenience for customers. They also realized they could service even small Internet orders at little or no additional cost to REI or the customer if the customer was willing to wait and pick the shipment up at the store. Because these shipments ride with regular less-than-truckload (LTL) store-replenishment orders, the cost is literally pennies per pound.

Similarly, REI managers were smart enough to offer free shipping on very large items such as kayaks, but only if a customer service person lets the customer know when the shipment can travel with other items on the periodic truckload replenishment shipments from the central warehouse. Finally, REI has installed Internet kiosks in its stores, allowing customers to order out-of-stock items directly. In each case, REI's logistics professionals have participated strongly in those customer-fulfillment efforts, as related in the discussion on kiosks in next section.

One of the primary appeals of online shopping is the nearly limitless product selection. But if customers cannot find something or have to wait, the advantage of such virtual inventory is nullified. From the online consumer's perspective, there is less of an excuse for being "out-of stock." In cyberspace, Internet retailers can draw on a wide-reaching supply base. At the same time, there is usually little to stop consumers from shopping elsewhere if they do not find what they want on their first-choice site.¹

It is up to logistics managers, supported by smart search routines and fast Web sites, to create the perception that virtual shelves are always full. This is a challenge to their inventory management skills; they have to scale up existing capabilities and take on new roles as the custodians of the virtual shelves. In practice, they have to draw not only on their own inventory but also on distributors' inventories and even on partners' stock, without exposing the customer to all the details.

The combination of scale and the highly distributed nature of online inventory requires a continuous search for patterns, rules, and other regularities that can save time and eliminate errors. In order to accomplish this, data mining, neural network pattern recognition, and other advanced data management techniques will have to be blended with inventory management systems. Our studies found several key points at which logistics expertise can add value, particularly with regard to inventory segmentation and product information.

Inventory Segmentation by Profit Potential

Logistics managers need to pay particular attention to inventory that generates a disproportionate share of overall profits. Specifically, they have to become clever at segmenting inventory, working with marketing to provide product information (such as shelf inventory turns), and overseeing linkages with partners that hold key inventory. Yet that information is often not shared and may not even be compiled.

We found very clear evidence of the need for effective segmentation when we broke down our margin data by key CD characteristics. Our survey data revealed that less popular CDs carried a higher margin, at more than 30 percent, than more popular CDs at 20 percent. Companies need to be careful, however, that poor logistics practices don't erode this margin. Some retailers understand that, and they have developed e-commerce sites that specialize in areas and items that have less competition and thus can support higher margins. Their logistics managers know that the toughest inventory task is stocking a new product with no history, and sparse demand only makes this job even harder. To succeed, they have to know where they can locate obscure, low-demand items.

Scanning for such inventory requires logistics managers to be sensitive to the world beyond their company's stock. Locating and connecting with the huge number of sites that might be useful is a task outside of the logistics manager's traditional domain. Even if the sourcing professionals can find and qualify special-purpose sites, it becomes logistics' responsibility to keep track of the inventory, policies, and continued existence of those important sources of higher profits.

Segmentation in Practice

Fortunately, spreading risk is much easier on the Internet than in physical retailing, if the supply chain is structured and operated carefully. Leading Internet retailers have taken several approaches to managing remote inventory and inventory that they do not own. What follows are three examples of methods to tackle the problem of managing worldwide virtual inventory that must be immediately visible and available.

1. Trans-Shipment. Consumers on the Internet do not know (and need not care) where their purchases are physically located. E-retailers and their suppliers can take advantage of that fact to minimize risk by centralizing inventories in only a few wholesaler sites and trans-shipping product between sites as necessary. It's thus easier for e-tailers to centralize inventory because e-retailing provides a single point of contact with shoppers—the Web site—as opposed to physical retailers with many different, individual stores. These orders can then be made easily visible to the e-retailer's supporting vendors. The large wholesalers see consolidated demand across multiple retailers, which is inherently more stable than for individual stores or retailers. For less popular CDs, for example, the ability to consolidate demand yields better availability for all retailers offering these CDs on their sites.

However, the Internet has not eliminated the importance of time and distance. Logistics managers still have to analyze the trade-off between inventory and transportation costs and determine where to put inventory. Similarly, cycle time depends on both inventory availability and distance to the customer. True competitive advantage is achieved when logistics managers can assure short cycle times without increasing inventory costs for themselves or their suppliers. The ability to trans-ship between vendors' inventory facilities has proven to be especially helpful here. Sharing inventories is relatively easy when communication is virtually instantaneous and physical location is not an issue for customers.

2. Multi-Layer Inventory. Amazon.com, arguably the most successful pure-play Internet retailer, has devised a classic three-echelon process for stocking its virtual shelves. As with any inventory system, the fastest-moving CDs are generally stocked closest to the customer—in this case, the Amazon-owned and -operated distribution centers. If, for example, a CD is not available at Amazon's own warehouse (none of the CDs we ordered were), then a major distributor is queried and instructed to ship the product in an Amazon.com box directly to the customer. If the distributor is out of stock, Amazon can pass the order immediately to the manufacturer, which may release its inventory on demand.

Relying on such multilayered inventory for CDs and many other product categories has contributed to Amazon's huge returns in asset productivity. Compare the numbers: For the first nine months of 2003, Wal-Mart used $77.5 billion in fixed assets and inventory to generate $182 billion in sales. Yet Amazon used $451 million in fixed assets and inventory to generate $3.3 billion in sales, making its asset productivity, based on sales, three times better than Wal-Mart's.

3. Kiosks. A number of brick-and-click retailers have installed Internet kiosks (effectively customized terminals) in their stores to give customers and store personnel visibility into the retailer's Web site and inventory data.² This hybrid model promotes virtual sharing of inventory between stores and allows the logistics manager to react more quickly to shelves that have emptied anywhere in the physical network. The number of in-store kiosks is projected to grow at a 17-percent annual rate through 2006.

REI, Borders, The Gap, and Men's Wearhouse are among the retailers that have implemented this model. The advantages are many. Because customers can now find out if out-of-stock items are available at another of the retailer's stores and can order any of the 78,000 items REI carries (twice as many as are available at REI's largest stores), they're more likely to place the order then and there. Additionally, store employees can order items for customers and provide information from the Web site using their cash registers, which have been Internet-enabled since 1999.

While home delivery is usually an option, many customers opt to pick up the items in the store or at another one nearby. In those cases, there is no additional delivery cost for the retailer because the items join regular store-replenishment shipments.

REI attributes increased sales to the presence of kiosks that provide inventory and other information. Similarly, at Borders, special orders are up as much as 20 percent at some stores.

Kiosks effectively allow individual salespeople and customers to put backorder and trans-shipment requests into the logistics system. Once again, the scale of the logistics job increases as the number of transactions multiplies. It is up to the logistics manager to design systems and procedures that will cope with the increased variety of options— such as transfer between stores, hold at alternate location, and ship to store with replenishment. Trans-shipment and backorder fill at the store level are not new logistics activities, of course. But handling those tasks in full view of the customer may well require consumer-relations skills that have never been part of most logistics organizations.

The Logistics of Product Information

In physical stores, customers can fully experience the product before deciding to buy. They can see if it is available, inspect the packaging, read the labels, even play a track or two of a CD or read a few pages of a book. The ability to sample products has become so important that experts in consumer behavior now include a preview phase in models of the buying process.

As custodians of the virtual shelves, logistics professionals must now provide information that replaces the sensory experience of seeing and touching the product. This information may go far beyond the usual inventory balances for which logistics professionals are typically responsible. Dimensions, sizes, available colors, and even health information may have to be part of the product offering on the site. There are no store personnel to update prices, organize assortments and facings, or to monitor product turns. As logistics already has to get availability and product characteristics correct for inventory-management purposes, it makes sense for it to collaborate with marketing and take on some of the merchandising functions of the virtual store.

The information itself is a key component of the e-retailer's profitability. In our study, we found a nearly perfect correlation between information provided and product margins. (See Exhibit 3.) E-retailers of CDs routinely provide sound clips as well as industry expert reviews, although the amount of information varies by retailer. We constructed an index of product information availability based on the number of reviews and sound clips available for each CD purchased, and then combined purchases by retailer. Internet retailers that provide more product information believe customers will buy more frequently and pay higher prices. That accords with the experience of REI, which believes that the information provided by in-store kiosks is partly responsible for an increase in sales that is equivalent to the annual revenue from one of its 25,000 square-foot stores.

The availability and completeness of product information must be tracked just like any other product attribute, adding to the inventory-management task. That's not a foreign problem for logistics professionals; in some companies, logistics is responsible for maintaining the master database for raw materials and finished goods, including all relevant dimensions, environmental limits, and packaging requirements. Pricing issues are also familiar territory for logistics organizations that have shipping and customer service under their umbrellas. Customization and postponement inevitably require familiarity with the details of product structures. In summary, logistics professionals look like good candidates to keep vital information up to date on their retailers' Web sites—especially when they collaborate with the product managers.

The Internet retailer needs to pay attention to certain cost and service issues that physical retailers can ignore.³ The cost side is relatively complicated. Internet sites typically position shipping and handling (S&H) charges as add-ons to the purchase price. Customers have to be convinced that the delay and expense of home delivery, which they perceive as an inconvenience (at best) or a surcharge (at worst), are acceptable aspects of Internet buying. Such charges are the main reason that many online customers abandon their shopping carts.⁴

Recognizing this, Amazon.com, for example, has used S&H charges for promotional purposes with spectacular success. After Amazon.com lowered its order limit for free fulfillment from $99 to $25, economies of scale and scope grew. As a result, internal fulfillment costs dropped from 12.8 percent to 10.6 percent of sales.

On the service side, our study tracked the service for each of our purchases from order placement to final delivery. Specifically, we recorded the times from order to acknowledgement, order to shipment, and shipment to delivery. The results are shown in Exhibit 4, where column 3 is the average time from order to delivery for each retailer. We were surprised to find that the retailers we tested fell into two distinct groups, based on service promised and service delivered. The finding reveals a significant "profit hole" that logistics professionals could do much to fill.

One group of retailers—the top table—promises approximately nine-day delivery cycles and fulfills that promise. This group promises leadtimes of eight to 10 days and delivers within half a day of that promise. We labeled them the "deliver as promised" (DP) group. Their average margin is more than 29 percent.

The other group delivers early—well within promised leadtimes that range from less than seven days to 18 days. (The largest discrepancy was for Alldirect, which delivered on average in less than half its promised leadtime. FYE showed the smallest gap; the retailer took roughly 75 percent of its promised leadtime to get products to customers.) We named this set the "under promise/over deliver" (UO) group. Examining the two elements of direct fulfillment—order receipt to order shipment and order shipment to final delivery—we find that both DP and UO retailers are essentially identical on the second metric. But the UO group excels in accessing shippable inventory. Their logistics managers have done a superb job of streamlining processes and interfaces with their key distributors.

However, the UO retailers have comparable margins to the DP group (except for Alldirect, which has an even smaller margin than the DP retailers). This fact indicates that they have gained no benefit from their superior performance of faster delivery than their own promises and faster delivery than the DP retailers. The UO retailers don't seem to understand the potential competitive advantage of their logistics networks. They have two choices, either of which should lead to improved performance.

• Slow down, deliver later. The UO retailers could stop delivering early, allowing their logistics staff to use cheaper transportation. Shipments could be consolidated and given to "zone skip" companies to cut down on parcel delivery expenses, low-cost carriers could be utilized instead of premium operators, and in some cases, intermodal service could replace standard parcel delivery.
• Stay fast, and brag about the better service. They could exploit their promises to be competitive with or superior to the DP retailers. That would immediately help them reposition themselves as high-service operations. UO retailers such as Walmart.com and Bestbuy.com could then get the best of both worlds. Based on the offline branding of these retailers, consumers know that prices will be low at those sites. But tighter service standards would signal that low price does not mean low service. That unbeatable value proposition should encourage repeat business with no additional cost.

Logistics managers at the under promise/over deliver retailers should stand up and be counted. They are delivering a superior service, but their management, perhaps intent on projecting a low-cost, bare-bones image, is conditioning customers to low expectations. If logistics managers promote the superior value customers are receiving, then the UO group, already as profitable as its less service-oriented competitors, can be further rewarded for its superior service.

We suggest logistics managers duplicate our study on a smaller scale and see which group they fall into, if they do not know already. Then they can take the results to their company marketing managers and begin discussions of service strategy. Failing that, they can use our results to show that in some companies the promises are already being far outstripped, and standards can be tightened without disappointing customers. Ideally, appropriate adjustments in service standards and/or pricing should follow.

Internet retailing is still about selling the right stuff online to customers who are willing to pay a fair price, and then taking care of those customers over time. But eliminating physical stores also brings logistics front and center on both service and cost dimensions. We believe that e-retailers' logistics managers already have much more influence than they are credited with. All the evidence points to an increase in that influence as Internet retailing gathers force. The findings from our study confirm that logistics professionals must play much more active roles in both the strategies and operations of their Internet initiatives.

On one level, they already command rich sets of competencies that can add value. On another level, they have much to learn from their increasing exposure to the interaction with the customer. As such, they will need to collaborate more extensively with other functions—particularly with marketing, customer service, and product development—broadening their spheres of activity from a conventional operational focus on transportation and warehousing to a more strategic emphasis on customer value and asset utilization. For many logistics professionals, this calls for a dramatic shift in self-image.

Tactically, they must pay attention to service promises and performance to minimize costly mismatches on the Web. They now must move beyond merely maintaining their core responsibility for basic inventory status information to also assuming ownership for much more detailed product information and the integrity of that data.

There is also much they can do to help minimize fulfillment charges—charges that account for half of total operating expenses at Amazon and that are the primary reason why some shoppers avoid Internet retailers. Logistics managers should use their expertise to carefully analyze choices around trans-shipment and backorder policies, promotions, and shipping and handling charges.

Strategically, logistics managers deserve to take a seat at the table when the online business model is designed. Right now, brick-and-click retailers seem to dominate much of cyberspace, and it is up to logistics to show the company how to share operating assets and expertise between the offline and online operations.

In any event, all signs point to the continued growth in e-tailing, and the characteristics that create success are becoming increasingly evident. Sector leaders are beginning to emerge. The retailers that fail to capitalize on logistics' leverage will not be among them.

Authors' note: This work was supported by the W. P. Carey School of Business and the Center for Services Leadership at Arizona State University. The authors thank John Lovett at Gomez.com for his support in the data-collection process. Thanks also to seminar participants at the INFORMS 2002 Annual Meeting and Joseph P. Bailey at the University of Maryland for helpful comments on this article. Rajiv Sinha's research was supported by the Dean's Award for Excellence Summer Grant Program funded by the Dean's Council of 100, The Economic Club of Phoenix, and the alumni of the W. P. Carey School of Business.