3 Critical Issues in Internet Retailing

In the early, hyped-up days of e-commerce, Internet retailers tried to focus customer and investor attention on the bells and whistles of their product offering or Web pages, and hoped that no one noticed the poor performance of backroom operations — or they deluded themselves into believing that good execution was unimportant. But that approach resulted in late shipments and bloated fulfillment costs, which led to the demise of erstwhile leaders such as Pets.com, Webvan and Value America. In contrast, successes like Amazon.com invested heavily in building operations capabilities rather than outsourcing anything that didn’t appear sexy enough for the “new economy.”

Internet retailing now has completed the cycle from overhyped promises to overreactive retrenchment and has settled into a steady but heady growth pattern. Online retail sales in the United States exceeded $85 billion in 2005, and in 2006 they appeared to be on track to grow at around 24%.1 This would mark the fifth straight year at that rate, which is more than three times the growth rate of total retail sales.

With e-commerce sales now accounting for large shares of seven retail categories — over 40% in computer hardware and software and around 17% of music and video sales2 — consumers have become increasingly pragmatic in their purchasing decisions. This sustained increase in Internet sales has been accompanied by an increased focus on operations execution.

The growth in online retail sales can be explained not only by the emergence of “pure-play” Internet retailers but also, to a large extent, by the entrance of multi-channel players — traditional brick-and-mortar retailers leveraging their brands and hopefully their physical infrastructures to compete on the Web. “Big-box” retailers such as Office Depot and Staples each generated more than $3 billion in Web sales in 2005. And even 114-year-old Sears ranks among the top 10 Internet retailers (largely due to its 2002 acquisition of cataloger Lands’ End), according to Internet Retailer magazine.

With the increasing maturation of online commerce, many Internet retailers now face important inflection points in their operations strategy. Managing Internet retail operations will increasingly involve more than simply managing costs. It will require an understanding of the unique challenges of the category and target market — as well as the individual product and customer. Each of these factors produces a different cost-service tradeoff and, accordingly, different “right” answers. More complex still, the answers will keep changing as the online channel grows, customer expectations evolve and operational options expand.

Going forward, there are three critical operations issues facing these retailers.

1. How should returns be managed to achieve immediate customer profitability and long-term loyalty?
2. Is the structure of the physical distribution network optimal?
3. Where should product inventories be deployed across the network for the best cost and service combination?

About the Research »

Manage Returns to Enhance Loyalty and Profits

Returns occur more often and thus play a more critical role in Internet retailing than in traditional brick-and-mortar retail. Some “commodity” or “hard-good” items, such as books or CDs are fairly easy to search for and examine online and, therefore, do not require liberal return policies for customers to feel confident about their purchases.3 However, since customers generally cannot examine online purchases until after the item has been paid for and delivered, return rates can run as high as 30% for some Internet retailers selling specialty items, such as apparel, and many currently feel compelled to offer free returns to compete with traditional retail options.

Many companies have turned to specialists to handle the overwhelming returns problem. For example, returns management firm Newgistics Inc., headquartered in Austin, Texas, and customer support specialists Global Response, headquartered in Margate, Florida, provide networks of facilities and an infrastructure for “reverse logistics” — that is, the shipment of goods back to the source. By focusing on their specific tasks and sharing facilities across multiple clients, these specialists offer faster execution and economies of scale. But crediting the customer and recovering the goods still leaves open the question of whether to return the goods to stock or to dispose of them through alternative, discount channels. Online sellers with offline retail presence, such as electronics retailer Best Buy Co. Inc., have an advantage in this regard over pure-play online retailers. They can easily leverage their existing infrastructure to offer customers the potentially more convenient option of returning online purchases to the brick-and-mortar stores, while affording themselves the often cost-saving option to aggregate returns and send them back in bulk to regional or national distribution centers for restocking, refurbishing or scrapping.

Ephemeral demand and potentially sullied products present obvious operational challenges for Internet retailers, but the handling of customer transactions has much broader consequences. Our recent survey of more than 400 customers of five Internet retailers covering a range of product categories found a strong, positive link between both the returns transaction and the returns policy and customer loyalty. Customers who spent less time in preparing their returns and waited shorter amounts of time to receive their refunds indicated far more satisfaction and loyalty than those who had to take more time and wait longer. (See “The Relationship Between Customer Time and Satisfaction.”)

The study also underscored the point that prior service experience shapes the customer’s expectation, reinforcing the importance of repeatable operations excellence in maintaining customers.4 Specialty retailer Frederick’s of Hollywood Inc., headquartered in Los Angeles, California, understands this.

In addition to a standard returns process, it offers customers a premium service for returning Web site purchases through Newgistics and other outside service providers, which minimizes consumer effort in preparing the return and expedites refund processing.

Order execution is another important way for Internet retailers to improve returns management. The best place to prevent a return is at the point of purchase. For instance, eBags Inc. — the leading online retailer of luggage, briefcases, backpacks and handbags, based in Greenwood Village, Colorado — continually expands its product views and provides more detailed product specifications and customer usage ratings to improve the initial selection decision.

Besides improved execution, structural factors must be taken into consideration. Our ongoing research with a consumer durable goods retailer has found that the more expensive an item, the greater the likelihood it will be returned. Just as the effort of preparing a return influences customer loyalty, the money at stake also affects the likelihood that a customer will bother to mail back the item. Initial research also suggests that for a given price level, a customer is less likely to return a bulky item than a smaller, easy-to-handle product.

Less intuitively, goods with lower sales have higher return rates. That is, after controlling for other factors such as price and size, we found that customers return popular items less often than they do items that sell infrequently. For retailers, this effect represents a form of double jeopardy in which the most returns occur with the products least likely to be resold, thereby generating higher average inventories and cost-to-serve.5

This preliminary finding offers contradictory evidence to the emerging theory of “the long tail,” which suggests that Internet retailers can profit by pursuing the fragmenting tastes of the consumer base and by offering a broader range of products.6 If more options enable greater “fit” for consumers, why do consumers return the less popular items more frequently after inspecting them? And more critical still, do returns and potentially other handling costs make these items even less profitable for retailers?

Further, return rates drop the longer an item is offered on a Web site (holding constant other product characteristics, such as price and popularity). This effect may be partially explained by the systematic discontinuation of low-selling items. But it also raises questions about the appropriate level of “churning” — the common retail practice of changing product offerings frequently. New items may attract more customers, but they come with a higher risk of return than older products.

Companies clearly need to understand product characteristics that drive returns, but they also should identify what we call “devil customers.” One company had regular customers who would order $10 worth of products they wanted, plus another $40 of undesired products so they could obtain free delivery for a $50 order. These customers would then return the $40 of unwanted products. Because the vast majority of customers behave fairly and reasonably, companies can structure return policies to allow their customer service representatives to go the extra mile for honest customers while stopping such egregious abuses.

Moreover, “devil customers” who take advantage of promotions and low-price deals prove fairly insensitive to the time needed for refunds or replacements. So even when approving returns from these customers, savvy Internet retailers can better manage cash flow and profits by giving priority to the customers who contribute to higher margins and show greater loyalty.

In sum, our research points to five principles that will continue to drive better returns management.

1. Keep the interaction simple to minimize customer effort. Customer effort is the single most important driver of returns satisfaction.
2. Provide multiple options for returns where possible. Some customers are happy to perform the transaction completely online, while others want to talk to a human operator.
3. Communicate clearly to set expectations. Customer expectations cannot be exceeded unless the company first sets them, then keeps the customer apprised in order to achieve a “no surprises” relationship.
4. Measure and manage. Track returns to identify the problem products and the “devil customers.” Neither can be addressed unless it is first identified.
5. Operate consistently and treat customers fairly. They will reward you.

Develop an Optimal Network Structure

Internet retailers have structured their physical distribution networks in a wide variety of ways, primarily as a function of strategic intent and scale. Some develop distribution competencies in-house, while others depend heavily on outside service providers. The number and geographic location of fulfillment centers also varies, from a single location driven by happenstance to multifacility, global networks strategically designed to optimize cost and service.

Andrew Westlund, former vice president of Global Operations for Amazon. com Inc., explained that company’s early decision to vertically integrate into physical distribution by arguing that no fulfillment vendors in the late 1990s really knew how to manage Internet fulfillment operations.7 Amazon’s strategic decision to integrate forward vertically and commit to operations excellence allowed the company to maintain customer service levels during the 1999 holiday season while other Internet retailers experienced meltdowns. During the following years, Amazon, which is headquartered in Seattle, Washington, closed and opened facilities in a continual quest to find the best footprint, evolving from a single backroom operation to a network that currently encompasses 17 fulfillment centers within the United States and another half-dozen facilities outside the United States.

Several different considerations can drive a company to move from a single facility to multiple sites. Most obviously, as an Internet retailer expands its geographic scope, the resulting need to lower transportation costs and speed delivery can drive it to open additional facilities that are closer to customers. Less obviously, product characteristics can influence the network structure. Small items can be shipped cheaply even over long distances, while outbound delivery of large, bulky items costs more. Complicating the decision making is that large items are more likely to achieve good inbound shipping economies, even to multiple facilities. All these issues need to be examined by retailers in the context of their business, products and demand characteristics (see “Tradeoffs in Designing a Distribution Network”).8

Without the funds to build their own networks, smaller firms have outsourced fulfillment, thereby converting significant fixed costs to a variable cost. Finding a capable outsource provider can be difficult when the product requires specialized handling, however. For example, pet pharmacy 1-800-Pet-Meds, owned by PetMed Express Inc., operates out of a single, in-house fulfillment center in Pompano Beach, Florida, far from most large centers of population in the United States. Although such a location would produce a significant cost disadvantage in many retail categories, it works for 1-800-PetMeds because most of the company’s shipments are sent via U.S. Postal Service Priority Mail, which is priced at a flat rate regardless of the shipping destination. The special staffing and controls required for the handling of prescription medicines also pushed 1-800-PetMeds to keep fulfillment in-house.

At the other extreme of distribution networks, eBags avoids the backward integration model and instead asks its original equipment manufacturer suppliers to provide fulfillment service through drop shipping. The luggage suppliers that provide products to eBags, such as Samsonite Corp., already operate multiple fulfillment centers dispersed around the country. Luggage producers typically serve small luggage retailers via United Parcel Service shipments (unlike many other manufacturers’ distribution centers, which handle only full truckload shipments). For this reason, OEMs can offer drop shipping at a lower cost and better service level than eBags could do on its own. This service has allowed eBags to operate only a minimal fulfillment infrastructure for some select (mostly private-label) product lines and instead to focus on integrating its suppliers through advanced information systems, including a sophisticated package tracking system and supplier performance reporting process.

The choices for a traditional retailer with a growing online presence prove no less daunting. Consider Clifton, New Jersey–based Linens ’n Things Inc., the second-largest retailer of home goods in the United States. Despite operating three distribution centers to serve its network of over 500 stores across North America, the company chose to outsource fulfillment operations of its $70 million online operation for LNT.com to .Com Distribution Corp., which operates a 300,000-square-foot facility in Edison, New Jersey. The outsourced facility, which LNT.com shares with other customers, offers economies of scale while allowing Linens ’n Things management to focus on the larger business and their core merchandising capability.

With such a wide variety of options, the appropriate network architecture is anything but obvious. Furthermore, the optimal answer changes over time as online retail sales grow in importance. Consider the example of Borders Group Inc., a $4 billion retailer of books headquartered in Ann Arbor, Michigan. In April 2001, Borders outsourced its Web site and fulfillment operations for Borders.com to Amazon. The difficulties of managing a separate online channel seemed disproportionate to the relatively small revenue it generated as an overall percentage of sales (less than 1% at the time of the agreement). However, strategic dynamics change over time. Borders is seeking to increase the use of interactive marketing via the Internet, but it finds the formerly logical outsourcing decision a constraint. Although still cost-effective, the current model limits Borders’ ability to capture the customer directly online.

Despite these examples, our research shows that, in many cases, the supplementary benefits offered by external partners prove very attractive to Internet retailers. These benefits normally translate not only into broader fulfillment network footprints but also into more complete and reliable fulfillment services.9 For example, outdoor-gear retailer Recreational Equipment Inc. has partnered with external providers to manage customers’ requests and inquiries related to in-store pickups of online orders. This has allowed REI to improve its in-store inventory utilization by better matching stock availability with demand.

Clearly, no “one-size-fits-all” network structure applies to Internet retailers. What’s more, the “right” answer will keep changing if Internet retail sales continue to grow at 25% per year. The decision ranks among the most critical of strategic issues, however, due to the capital implications of vertical integration and the “lock-in” risks of picking the wrong network partners. The best Internet retailers will invest in a network architecture that fits their specific product characteristics and strategic competencies at a price they can afford.

Deploy Inventory to Profitably Meet Customer Needs

As their distribution networks grow larger, pure-play online retailers will face increasingly complicated decisions regarding inventory deployment. Our recent study of online retailing profitability showed that excellent purchase fulfillment generates larger gross margin. In the case of one particular Internet retailer, a one-day reduction in delivery lead time implied an 8% to 18% improvement in gross margin.10 Achieving more rapid delivery depends upon inventory availability — not only in terms of quantity but also in terms of where such inventory exists within the supply network.

Although Internet retailing purports to offer a “virtual” inventory model unconstrained by the limits of a physical store, for fast delivery the product needs to be stored somewhere by someone along the supply chain. Amazon.com, for example, offers more than 10 million book titles, 50 times that of a typical Borders superstore and 400 times that of a mall-based bookstore. But Amazon fulfillment centers carry only a fraction of the titles offered; wholesalers hold many additional titles for rapid delivery to Amazon, while other titles must be sourced directly from the publishers. Some out-of-print titles can be sourced only from used or closeout specialist bookstores.

To decide whether to hold a book in-house or to order as needed from a distributor, Amazon must do more than calculate the cost of carrying a book under each option. The company must also consider whether to stock the book at a single facility or to duplicate inventory across multiple locations. Further complicating matters, Amazon must consider the impact of complementary or bundled products in every order, since “split shipments” rank as one of its most critical operational cost drivers. For example, it may minimize delivery costs by shipping a single package with two items from Nevada to Georgia rather than shipping one package from Delaware and another from Kentucky. Although not a normal practice, Amazon could also delay the order and make a bulk trans-shipment between two facilities.

Our own research on inventory deployment for Internet retailers has highlighted these tradeoffs in supply chains comprising “speculative,” forward-deployed inventory and “postponed” stocks held at supplier locations rather than in-house. We found that a 50% increase in transshipments raises average transportation costs per order by 21% but reduces average back-order time by 8.5 hours.11 Good service, including in-stock availability, proves paramount for sellers of commodity products such as CDs because consumers can easily divert their purchases to another retailer with better availability. Outsourcing considerations also enter into the equation. Aligning incentives along the supply chain can reinforce delivery performance and perhaps allow even better promises in subsequent orders.

Inventory deployment presents greater challenges to multichannel retailers, which offer products through both a Web site and physical stores: What can or should be made available exclusively online, at the store only or both, and how to price accordingly? Poor inventory visibility presents one of the greatest barriers to the effective integration of offline and online retail operations. Research on retail operations execution has highlighted the poor understanding that most retailers have of actual inventory levels in a given store. One rigorous analysis of a sophisticated, multibillion-dollar chain found inaccurate records for 65% of the stocked items, with an average error of 35%.12

For multichannel retailers, where to optimally deploy inventory naturally depends on fundamental issues such as the network architecture and degree of outsourcing. But beyond those system design considerations, these retailers should take into account the salient characteristics of the products being sold and the customers buying them. Consider how product characteristics determine electronics retailer Circuit City Stores Inc.’s cost-to-serve and optimal inventory deployments.13 Circuit City can gain significant transportation savings by shipping a bulky item, such as a wide-screen television, in full truckload shipments through its nine regional distribution centers serving its 621 domestic stores — whether a customer orders it online or at the store. At the other extreme, the most expensive camcorders weigh less than a couple of pounds and accordingly can be shipped by UPS ground service relatively cheaply.

Furthermore, if the camcorder has low unit sales, Circuit City might decide against stocking it in the stores at all due to the inventory carrying cost, including the risk of obsolescence. But a “pooled” central inventory offered exclusively over the Web site and held in a central distribution center might be justified, since far fewer units of inventory would be required.

Brick-and-mortar retailers can find a great advantage in the combination of Web sales and in-store pickup. Drugstore chain Walgreen Co., for example, allows customers to use its Web-based prescription ordering tools to schedule medication refills and pickups at a store of their choice among thousands across the United States. The advantage for Walgreen’s customers comes from the ability to schedule and track their prescription refills. Although clearly a customer convenience, the scheduling feature has also given Walgreen’s a powerful tool to forecast a category that accounts for more than 60% of its annual sales and better plan the prescription processing work-flow at its stores. This has led to better inventory management and personnel utilization, as well as fewer customers waiting for their prescriptions at the stores.

Although optimal inventory deployment must start with an understanding of customer preferences, it need not stop there. Customer behaviors can be changed with a compelling alternative value proposition. Often, the key comes from breaking traditional constraints, such as offering low cost and high quality, or speed and variety. The virtual nature of Internet retailing enables new possibilities, just as the introduction of big-box retailing 30 years ago presented consumers with a fundamentally new combination of low cost and high variety. Savvy Internet retailers are focusing on the unique opportunities of the online channel to create new service models that offer a superior value proposition to their targeted consumers. But most importantly, optimizing operational performance in Internet retailing will require understanding and managing the details.