the marketing and logistics efficacy of online sales channels

Upload: suryanarayanamurthy

Post on 11-Oct-2015

23 views

Category:

Documents


1 download

DESCRIPTION

logistic efficency

TRANSCRIPT

  • The marketing and logisticsefficacy of online sales channels

    Shashank Rao, Thomas J. Goldsby and Deepak IyengarSchool of Management, Gatton College of Business and Economics,

    University of Kentucky, Lexington, Kentucky, USA

    Abstract

    Purpose The purpose of this study is to investigate key differences between web-only and multi-channelretailers in terms of five different measures of web activity and three different forms of outsourcing behavior.Specifically, the research examines the marketing and logistics efficacy between business-to-consumer(B2C) retailers who sell exclusively via the web and retailers for whom the web offers one additional channelfor sales. Finally, it is suggested that how this study may give rise to future research in this area.

    Design/methodology/approach This empirical study using the lens of transaction costeconomics (TCE) to examine hypotheses regarding customer buying behavior and the retailersproclivity to outsource is conducted. Secondary data sources provide key metrics for the more than 250companies found in the sample.

    Findings Several key differences exist between the efficacy of web-only and multi-channelretailers, which can be explained with the TCE framework. Both web-only and multi-channel retailersare found to exhibit respective advantages. Multi-channel retailers enjoy more web traffic and offermore items for the consumer, yet are disadvantaged in terms of ease of search and conversion rate(percentage of shoppers who actually buy). In addition, web-only retailers are more likely to outsourcethe functions of logistics, marketing, and customer support.

    Practical implications This study has value to researchers and practitioners in that it illustrateshow two of the most common types of retailing alternatives differ from each other. Multi-channelretailers are challenged with the broad scope and immense collection of goods they offer and, therefore,struggle to convert shoppers into buyers. Web-only retailers, on the other hand, enjoy less web traffic,but prove more effective in conversion rates, perhaps related to their more extensive use of outsourcedexpertise in logistics, marketing, and customer support services.

    Originality/value In the decade since internet retailing (e-tailing) began to be accepted as a newsales channel, e-tailing has grown to a market size of over $160 billion within the USA alone. However,empirical examination of the functioning and performance of this sales channel is only nowcommencing.

    Keywords Electronic commerce, Internet, Marketing, Sales, United States of America

    Paper type Research paper

    1. IntroductionAs the reach of the internet has expanded at a substantial pace over the past few years,so too have the potential business opportunities that the internet supports. Fore-mostamong these opportunities is the sale and distribution of goods. For any company witha product to sell, the method of distribution is often as crucial a decision as developingthe product itself (Tsay and Agarwal, 2004). Early on, researchers had argued that the

    The current issue and full text archive of this journal is available at

    www.emeraldinsight.com/0960-0035.htm

    The authors would like to thank Karen Harmon for her assistance with this research, includingcollating, managing and preparing the data, without which this study would have beenimpossible.

    IJPDLM39,2

    106

    Received 4 August 2008Revised 20 October 2008Accepted 18 November 2008

    International Journal of PhysicalDistribution & Logistics ManagementVol. 39 No. 2, 2009pp. 106-130q Emerald Group Publishing Limited0960-0035DOI 10.1108/09600030910942386

  • internet would lead to fundamental changes in the way this sale and distribution ofgoods would be executed (Choudhury et al., 1998; Clark and Kauffman, 2000; Groveret al., 1999; Janssen and Sol, 2000). Central to this premise was the idea that comparedwith the conventional physical distribution channel, the internet could provide widercoverage with lower operational costs (Kumar and Ruan, 2006). The premise seems tobe holding true; several retailers and manufacturers have either opened online shops asan additional channel of distribution, or are seriously considering it (Stern et al., 1996;Bannon, 2000; Alptekinoglu and Tang, 2005; Tsay and Agarwal, 2004).

    The use of more than a single-channel to sell goods to consumers is calledmulti-channel retailing (Kim and Park, 2005). Multi-channel retailing could includephysical stores coupled with mail-catalogs to try to boost sales (e.g. J.C. Penney), orphysical stores coupled with online-stores (e.g. Circuit City, Staples), kiosks, or wirelesschannels (e.g. kayak.com), or a combination of the above (Clark, 1997). Researchers haveargued that with increasing competition, retailers tend to move towards channelproliferation (Gabrielsson et al., 2002; Alptekinoglu and Tang, 2005). Consequently, withgrowing competition in the marketplace and the increasing reach of the internet, severalretailers are moving from being purely single- to multi-channel, with the internetrepresenting the most common multi-channel complement to an existing physical storestrategy. As early as 2002, approximately 75 percent of brick and mortar retailers hadsome form of multi-channel retailing strategy either in place or in the works, to enhancethe firms overall performance (Kim and Park, 2005). Moving forward to 2005,approximately 40 percent of retailers sold through three or more channels, while another42 percent sold through two channels (Direct Marketing Association, 2005).

    Examples of firms successfully pursuing multi-channel retailing are plentiful; forexample, several computer companies such as Hewlett-Packard and Apple are adoptinga dual-distribution channel by not only selling through their traditional distributionchannel (e.g. Circuit City, Best Buy, Wal-Mart) but also selling products on their ownwebsites. Similarly Dell, which until recently was a pure make-to-order internet-basedseller has begun distributing through resellers with physical stores as well. This trend isseen amongst manufacturers and retailers not just in the computer industry, but acrossproduct categories. For example, as apparel-retailing competition got increasinglyfierce, Lands End, a direct apparel retailer famous for its mail-order catalog business,launched an online sales channel in 1995, and subsequently partnered with Sears to offerits products in the store channel as well in 2002 (Alptekinoglu and Tang, 2005, p. 803).

    This trend towards multi-channel retailing has not gone unnoticed by researchers.Attempts have been made to study the dynamics associated with multi-channelretailing. For example, researchers have suggested several reasons as to why firmsmove towards multi-channel retailing and also the strategic importance of partnershipsin such settings (Esper et al., 2003). One of the commonly suggested reasons for the moveto open multiple channels includes the suggestion that increased competition andthe consequent pressure on the top line are what act as catalysts towards increasedusage of multiple channels (Gabrielsson et al., 2002). Another popular line of reasoningimplies that recent advances in web-based information technology applications havehastened this trend (Alptekinoglu and Tang, 2005). Whatever the reasons offered for theproliferation of multi-channel retail, one key theme is clear: managers and researchersare recognizing multi-channel retailing as a new business paradigm that deservesspecial attention in the modern economy (Kim and Park, 2005).

    Efficacy of onlinesales channels

    107

  • Several avenues of research have been suggested in the field of multi-channelretailing. It has been noted that for firms that follow a multi-channel retailing strategy,channel management becomes increasingly complex as the firms must contend withissues concerning multi-channel pricing, branding, and consumer behavior (Alba et al.,1997; Brynjolfsson and Smith, 2000; Klein and Ford, 2003; Noble et al., 2005) all ofwhich are areas where lack of research is acutely felt. Similarly, Noble et al. (2005)argue that while consumer behavior in single-channel contexts has been studied insome detail, only recently have researchers begun to study the same in a multi-channelretailing context. Moreover, insofar as there appears to have been some research onchannel preferences and behavior (Kumar and Venkatesan, 2005; Montoya-Weiss et al.,2003; Ansari et al., 2008) the same cannot be said about the operational dynamics of amulti-channel retail structure.

    Consequently, research interest in the structure and functioning of multi-channelretail is increasing (Muthitacharoen et al., 2006). Among the most regularly researchedareas within the multi-channel retail format are those concerned with search costs(Rabinovich et al., 2003), channel conflict due to margins and price reactions (Tsay andAgarwal, 2004), order fulfillment (Rabinovich and Evers, 2003; Esper et al., 2003),channel choice (Kumar and Venkatesan, 2005; Montoya-Weiss et al., 2003), channelmigration (Ansari et al., 2008; Gensler et al., 2004) and pricing power (Tang and Xing,2001; Wu et al., 2008). However, scholarly inquiry into the differences between theoperational dynamics and performance of web-only retail and multi-channel retail islacking. The present paper is a step in this direction. Our substantive research questionis how do web-only and multi-channel retail formats differ in respect to theircustomer-facing and backroom operations? In particular, we look at differencesbetween web-only and multi-channel retailers based on front-end (customer-facing)measures such as store traffic, available product variety, ease of search, salesconversions, sales ticket sizes, and back-end factors such as the in/outsourcing ofsupport functions (logistics, marketing, and customer support). These are key variablesthat form the fundamentals of any retailing system (Chircu and Mahajan, 2006).

    We explore our research question in light of established theory that has gained wideacceptance in the study of channel structures and supply chain management:transaction cost economics (TCE). We demonstrate how differences in transactioncosts between web-only and multi-channel retailers explain differences in customerbuying behavior and the retailers proclivity to outsource across the two channel types.The rest of this paper is organized as follows: in Section 2, we present the conceptualdevelopment and hypotheses. The data sources and collection methodology arepresented in Section 3. Section 4 presents the data analysis methodology while Section5 presents the results of the study along with their implications and future researchavenues. Finally, Section 6 provides conclusions.

    2. Theoretical background and hypothesis developmentThis study is concerned with understanding two key factors in online retailingdynamics: channel structure and consumer choice. Channel structure refers tothe patterned aspects of relationships between (and within) channel participants(Devaraj et al., 2006, p. 1091). Researchers have traditionally used a TCE frameworkto study questions of structural analysis in channel relationships. TCE provides aparsimonious and robust methodology for addressing the issue of channel structures

    IJPDLM39,2

    108

  • (Bowersox and Cooper, 1992; Klein et al., 1990; Kohli et al., 2004; Wu et al., 2008).Moreover, it has been argued that not only does TCE help in addressing channelstructures, but it also helps explain consumer choice of channel structures (Liang andHuang, 1998; Sarkar et al., 1998). Consequently, it becomes valuable to study ourcentral research question using the established lens of TCE. In this section, we brieflyintroduce the major tenets of TCE and develop relevant hypotheses.

    Originally developed from the work of Coase (1937) on the efficiency of economicexchanges, TCE advances the idea that within an exchange involving buyers andsellers of a product, the market price of the product, while important, is not the solecriterion towards the purchase choice; indeed several other costs could exist that willbe considered by the buyer in arriving at this decision (Williamson, 1975, 1985).Williamson (1979, p. 261) argues that TCE is an interdisciplinary undertaking thatjoins economics with aspects of organizational theory and overlaps extensively withcontract law. Bowersox and Cooper (1992) identify six key influences of transactioncosts, three of which are economic and can be quantified, while the other three factorsintroduce behavioral influences that tend to serve as qualitative factors in the TCEframework. The non-economic factors include bounded rationality, opportunism, andenvironmental factors, while the economic factors include assembling information,bargaining, and performance monitoring.

    Search and information costs are the expenses which are incurred in determiningwhere the required product(s) is/are available and which seller has the most agreeableprice. Arguably, seeking alternatives that meet specific needs and comparingalternatives across product/service characteristics and cost usually involves more thanlistening to the sales pitch of prospective suppliers (Goldsby and Eckert, 2003). It iscommonplace to engage in substantial research of the suppliers offerings and the firmsthemselves; an activity that imposes a substantial unquantified cost on the consumer; acost that arguably would get implicitly transferred as a seller-specific expense.Bargaining costs are the costs required to come to an acceptable agreement with theother party to the transaction, including the devising of an appropriate contract orservice level agreement. Goldsby and Eckert (2003) argue that bargaining costs becomeimportant once the preferred alternative has been selected. Indeed, costs can rise withthe extent and means of negotiations. Policing and enforcement costs are the costs ofmaking sure the other party fulfills the terms of the contract, and taking appropriateaction, should they fail to fully comply (Bowersox and Cooper, 1992).

    2.1 Transaction costs and the buyerThe economic factors of TCE have clear and distinct impacts on behavior in differentchannels, an issue that has heretofore not been tested in online retail environments.Chircu and Mahajan (2006) describe the sequence of transaction steps that form theretail transaction chain: store access, search, evaluation, and selection, ordering,payment, order fulfillment, and after-sales service, each having its own associated costcomponent. We demonstrate below how distinct steps relate to the web-only versusmulti-channel retailing marketplace.

    2.1.1 Search costs and their channel impacts. Researchers have noted that among thevarious transaction costs, product search cost is one of the most important to consider(Bakos, 1997; Keeney, 1999; Chircu and Mahajan, 2006). Online storefronts often servenot only as transactional channels, but also as search channels for firms involved in

    Efficacy of onlinesales channels

    109

  • multi-channel retail (Kim and Park, 2005; Choudhury and Karahanna, 2008). It hasbeen noted that many customers routinely use multiple channels of the same retailer tomake purchases (Cleary, 2000; Forster, 2004). According to an industry report, 66percent of consumer respondents declared that they had made use of more than onechannel of the same retailer in the overall purchase (from product search to final receiptof goods) of items during their shopping experience (Saunders, 2002). In another study,Ratchford et al. (2003) showed that about 39 percent of respondents used the onlinechannel to get information before they purchased vehicles from dealers. Similarobservations suggest that the structure of markets in the future will adapt toaccommodate the search needs of consumers, along with their buying needs (Sharmaand Sheth, 2004). Based on these observations, it could be argued that online shopperscan be divided into the four categories as shown in Table I.

    Together, these arguments imply that web storefronts of multi-channel retailers mustcater to three distinct kinds of traffic (categories 1, 2, and 3); the first would use the onlinestore for the actual purchase (we call these the serious shoppers), while the second woulduse the online store merely as an information search medium and might make the finalpurchase from the brick-and-mortar store (we call these the window shoppers)[1]. The thirdwould use the offline store for search and the online store for the purchase. Meanwhile, thefourth category does not engage the internet for search or purchase activities. The noteddistinction between online window shopping (also called research shopping) and seriousshopping has received some discussion in the literature (Verhoef et al., 2007).

    As a result, multi-channel stores would receive a whole category of customers thatweb-only stores would either not receive or receive to a lesser extent (the researchshoppers). We therefore argue that multi-channel stores will have higher traffic densitycompared to that of web-only stores. This leads to our first hypothesis:

    H1. Multi-channel retailers will see a higher amount of web traffic as compared toweb-only retailers, after controlling for store size and product type[2].

    As noted, multi-channel retailers could be catering to two kinds of store visitors(window shoppers and serious shoppers) during search. Web-only retailers wouldtypically cater to a far smaller percentage of window shoppers. Consequently, it wouldseem plausible that multi-channel retailers would have to display a higher amount ofitems on the web site, in order to satisfy the greater variety of shopping behaviors thatvisitors to their websites would display. In order to remain competitive in the onlinemarketplace, multi-channel retailers would have to not only carry as many differentproduct categories as web-only retailers would carry on the online storefront, but oftentimes they would have to go beyond this minimal level and display information onstore-only products too. Some of the products displayed on the web site ofmulti-channel retailers may be available only in the brick-and-mortar store; a prospectthat would not apply to web-only retailers. This increase in product categories carried

    Channel for purchaseOnline Offline

    Channel for searchOnline Category 1 Category 2Offline Category 3 Category 4

    Table I.Types of shoppers basedon channel preferences

    IJPDLM39,2

    110

  • would manifest in greater variety and volume of stockkeeping units (SKUs) displayedon the web storefront. Consequently, we argue that multi-channel retailers would haveto display information on a higher number of SKUs as compared to web-only retailers.This leads to our second hypothesis:

    H2. Multi-channel retailers will display a higher number of SKUs on the web ascompared to web-only retailers, after controlling for store size and product type.

    2.1.2 Cost of evaluation and selection. Chircu and Mahajan (2006) argue that the cost ofevaluating and comparing across products is a second important transaction cost thatbuyers consider in a B2C context, prior to making purchases. We have argued in H2that multi-channel retailers will carry a higher number of available product options ontheir web storefronts as compared to web-only retailers, owing to the fact that they willbe dealing with a more diverse customer base than web-only retailers.

    It has been widely accepted that human beings have limited information processingcapacity, and consequently find it hard to accurately process all the information fed tothem (Simon, 1990). Researchers, in general, are of the consensus that increasing thenumber of options available to a shopper increases the complexity of thedecision-making process, thus increasing the cognitive load on the shopper (Bettmanet al., 1998; Lee and Lee, 2004). In keeping with these observations, we argue that whenonline shoppers are faced with a higher number of product options by way of anincreased number of SKUs on multi-channel retailers web storefronts, they willexperience a higher cognitive load in finding the exact product that they were lookingfor and also evaluating that product against the plethora of options available. Thisexpenditure of cognitive load will manifest in how difficult they perceive the task offinding the product that they were looking for when visiting a particular online store.This gives rise to our third hypothesis:

    H3. Multi-channel retailers will exhibit a lower ease of finding a particularproduct as compared to web-only retailers, after controlling for store size andproduct type.

    It has been argued that that online shoppers are more likely to purchase from a website when they are easily able to obtain and process product-related information fromthe web site (Lynch and Ariely, 2000; Zeithaml et al., 2002). In related research, Boyerand Hult (2005) demonstrate that web site ease is directly related to the extent ofpurchasing from the site.

    Based on our earlier hypotheses, we have suggested that customers would have tosift through a larger amount of information in a multi-channel stores web-interface,owing to the larger number of SKUs and the associated product information associatedwith each (per H2 and H3), thus increasing their cognitive load. On the other hand, theextent of sorting through information will be that much lower on the websites ofsingle-channel retailers, owing to the relatively lower number of SKUs associated withtheir web storefront. Arguably then, information evaluation and assessment would bemore difficult when shopping at multi-channel retailers than it would be whenshopping at web-only retailers. Thus, other things being equal, web site ease shouldbe lower for multi-channel retailers than it will be for web-only retailers.

    Linking these ideas with the findings of Lynch and Ariely (2000), we propose thatceteris paribus, the likelihood of visitors to an online store turning into actual buyers

    Efficacy of onlinesales channels

    111

  • will be less for a multi-channel retailers web site than it will be for a web-only retailersweb site. The likelihood of buying can be measured as the conversion rate, which isdefined as the percentage of visitors to an online store who buy the product from thatstore on that visit. Based on our above discussions, we hypothesize that:

    H4. Multi-channel retailers will have a lower conversion rate than web-onlyretailers, after controlling for store size and product type.

    2.1.3 Shipping costs and their impact. Within the domain of TCE, shipping costs andorder completion costs can be considered other examples of transaction costs. In astudy conducted across several internet retailers, Ancarani and Shankar (2004) foundthat the per-order shipping cost was higher for multi-channel retailers than it was forweb-only retailers. One explanation could be that multi-channel retailers are inclined tokeep their overall prices for online and offline products at the same level, so as to keeptheir offline customers happy and avoid channel cannibalization (Xing et al., 2004).It could be argued that if multi-channel retailers would raise the price level of theproduct itself in the online store, then they would lose online traffic upfront (Swinyardand Smith, 2003; Srinivasan et al., 2002; Swaminathan et al., 1999), which would bedetrimental to the store as a whole. In order to avoid this problem, they might be likelyto use a partitioned price mechanism, offering initial low prices on goods in order tocompete with the lower price of web-only retailers. Subsequently, however, they wouldequalize offline and online store price by adding higher shipping charges thanweb-only stores, thus bringing the total price of the online goods to a similar level asthat in the brick-and-mortar store.

    It has been argued that consumers do not evaluate the overall costs of apartitioned price as they would a single, consolidated price (Morwitz et al., 1998) andthus such a strategy has the potential to work well for multi-channel retailers.A competing explanation could be that the web is by its very nature a lower costchannel and, consequently, shipping costs for this channel can be lower than formulti-channel retail. This explanation however, is somewhat imprecise, since Laseteret al. (2006) demonstrate that often times the web may, in fact, turn out to be a moreexpensive channel than other channels.

    Whatever the reason for the differential shipping charges, one underlyingobservation is nevertheless quite clear; shipping charges for multi-channel retailersare, on the whole, higher than they are for web-only retailers on a per-order basis(Ancarani and Shankar, 2004). When online shoppers at the web storefront ofmulti-channel retailers face this higher average shipping charge on a per-order basis,they are likely to demonstrate a coping behavior by consolidating multiple purchasesinto one order, thus trying to save on high shipping charges. On the other hand, theshoppers at web-only stores will consolidate purchases to a lesser extent, given thatprices will incorporate only the variable costs of the units purchased, relatively lowershipping charges, and sales tax (which should be relatively constant across the twochannel alternatives). This difference should manifest in the average size of an orderfor web-only and multi-channel retailers. Thus, we argue that multi-channelretailers will see relatively larger order sizes as compared to web-only retailers.The average order size can be measured as the average ticket size, which is theaverage amount of money spent on each order at the particular store. This leads to ourfifth hypothesis:

    IJPDLM39,2

    112

  • H5. Multi-channel retailers will have a higher ticket size (average amount spent ona purchase) than web-only retailers, after controlling for store size andproduct type.

    2.2 Transaction cost theory and the sellerAs noted earlier, according to TCE, a buyer of a good will consider not only the cost ofthe product itself but also the sum of the transaction costs associated with a product ofinterest prior to making a purchase. Additionally, researchers have argued thattransaction costs (especially those associated with product returns) in online retailingcan be exceptionally high (Laseter et al., 2006, 2007). Consequently, online retailingfirms interested in competing successfully in the marketplace will be interested inreducing transaction costs associated with doing business with the customer. Onemethod by which firms try to bring down the transaction costs of activities isoutsourcing (Ang and Straub, 1998; Bolumole et al., 2007; Goldsby and Eckert, 2003).

    In simple terms, outsourcing refers to the deploying of an activity to an externalsupplier; and is an alternative to internal processing (Aubert et al., 2004). While manyassociate outsourcing with the production of goods, several other business functionsare commonly outsourced today (Arora and Forman, 2007; Aubert et al., 2004; Goldsbyand Eckert, 2003). Some of the business functions that are commonly outsourcedinclude logistics, marketing, and customer support functions (Ivanaj and Franzil,2006). The extent to which these functions will get outsourced may differ betweenweb-only and multi-channel retailers. The rationale for outsourcing each function isgiven below.

    2.2.1 Outsourcing the logistics function. A key element of transaction costs thatexplains outsourcing is asset specificity (Ang and Straub, 1998; Malone et al., 1987;Williamson, 1975). Assets can be machinery required to manufacture a product,knowledge needed to perform a service, or even an appropriate location convenient forengaging with the other party (Aubert et al., 2004). The degree of asset specificity is acrucial issue in the logistics domain (Ivanaj and Franzil, 2006). A high degree of assetspecificity reduces the profits of outsourcing and encourages the firm to organize thegiven activity in-house (Maltz, 1993, 1994; Hobbs, 1996; Bolumole et al., 2007). Pacheand Sauvage (1999) argue that within logistics, asset specificity corresponds to suchspecialized activities as warehousing for specialized products, packaging, andtransportation. Fundamental operations as transportation and warehousing sometimesrequire specific and costly investments such as refrigerated vehicles, deep-freezestoring surfaces for frozen foodstuffs, specialized forklift trucks, and guidancesystems, among others (Bienstock and Mentzer, 1999).

    It is likely that multi-channel retailers will already have available, at least to someextent, the facilities needed for such operations. For example, circuitcity.com hasessential infrastructure in the form of physical stores, warehouses, and an existingtransportations system on which the online retail channel can piggyback (Laseter et al.,2007) and, consequently, would have a lesser necessity to outsource the activitiesassociated with these assets. Web-only retailers, on the other hand, will typically nothave these facilities available, and will be more likely to seek outsourced services.This gives rise to our sixth hypothesis:

    H6. Multi-channel retailers will outsource the logistics function to a lesser extentthan web-only retailers.

    Efficacy of onlinesales channels

    113

  • 2.2.2 Outsourcing the marketing and customer support functions. Pisano (1990) hasargued that companies with limited experience in specific areas are better offoutsourcing these activities, since the supplier can then bring the knowledge investment(expertise) that the parent firm lacks. Web-only retailers, owing to their very nature,have limited contact with the customer as compared to multi-channel retailers.Multi-channel retailers experience the customer through multiple contact points (onlinestore, plus the brick-and-mortar store) while web-only retailers do so only through onecontact point (the online store), which typically involves no person-to-personengagement. Therefore, multi-channel retailers will have greater knowledge of thecustomer than web-only retailers. This contact and associated knowledge should thenmanifest in the way these two channels accommodate key customer-facing functions(target marketing and customer support). We propose that web-only retailers will try tomitigate this lack of customer knowledge by outsourcing these functions to specialistfirms, consistent with Pisanos (1990) premise. Moreover, web-only retailers should bemore open to outsourcing these functions since the risk of having an outside presencewill be perceived lower by these firms, who already experience very low customercontact. These ideas give rise to our seventh and eighth hypotheses:

    H7. Multi-channel retailers will outsource the target marketing function to a lesserextent than web-only retailers.

    H8. Multi-channel retailers will outsource the customer support function to alesser extent than web-only retailers.

    The research hypotheses are summarized in Table II.

    3. Data collectionThe sample firms represented in this study were selected from the internet retailer top 500guide (IR500) for 2007 and from Bizrate.com. The IR500, which ranks firms based on thevalue of their online sales, is one of the only comprehensive rankings of American onlineretailers (Brohan, 2008). The database lists the top 500 online retailers in the USA, withtotal sales exceeding $166 billion, accounting for over 60 percent of the total online retailingmarket (Howlett, 2008). It should be noted that we are comparing web-only retailers withmulti-channel retailers online B2C channel sales. In other words, the data reported belowconstitute the data from the web storefronts of the multi-channel retailers, not the data forthe overall company itself, thus enabling an appropriate comparison.

    To collect these data, IR500 researchers contact hundreds of retailers over asix-month duration. The starting point of the data gathering is the internet retailer

    Variable Hypothesis

    Web traffic H1: multi-channel . web onlySKUs on the web H2: multi-channel . web onlyEase of finding product H3: multi-channel , web onlyConversion rate H4: multi-channel , web onlyTicket size H5: multi-channel . web onlyOutsource logistics H6: multi-channel , web onlyOutsource target marketing H7: multi-channel , web onlyOutsource customer support H8: multi-channel , web only

    Table II.Summary of researchhypotheses

    IJPDLM39,2

    114

  • ranking, based on web traffic from comScore and Nielsen Online. The retailers areasked for data on store sales, channel type, product type, web-traffic, conversion rate,ticket size, and the extent of outsourcing of the relevant activities. In cases whereretailers do not provide data, these are estimated based on multiple interviews withindustry experts, and retailers are provided multiple opportunities to respond to theseestimates. The current research discarded cases that used these subjective data pointsin an effort to minimize the chances of measurement error and biases. Additionally,the current research discarded manufacturer data (e.g. nike.com, hersheys.com) sincethe aim of this study is to examine online retailers, not manufacturers. As a result,239 observations were eliminated on these bases, leaving us with 261 usable cases.

    3.1 Independent variablesSince we were interested in studying differences between web-only and multi-channelretailers, the type of retail channel (web-only vs multi-channel) served as the focalindependent variable. In addition to channel type, the control variables of store size andproduct type are independent as well. Each is described briefly.

    3.1.1 Channel type. The IR500 divides online retailers into four categories based onthe nature of their distribution channels: web-only, retail chains, cataloguers, andmanufacturers. Retail chains and cataloguers were combined to form the multi-channelretailers. As previously noted, manufacturers were dropped from the sample.Web-only retailers were coded as 0 and multi-channel retailers were coded as 1 andwere run as dummy variables in the data analysis.

    3.1.2 Store size. Online store size was measured as the total amount of online sales ofthe store through the web. The data were obtained directly from the company, asreported in the IR500 database. From these, Amazon.com was deleted because it was asignificant outlier. With $17 billion in online sales in 2007, Amazon.com aloneaccounted for over 15 percent of our overall online sales and, consequently, it wasconsidered prudent to drop it as an unrepresentative data point. Post hoc results,however, will illustrate how inclusion of Amazon.com affects the analysis.

    3.1.3 Product type. Researchers have argued that differences could be observed inthe dynamics of online stores based on the product category being sold in the store(Lynch and Ariely, 2000). Thus, we determined it necessary to control for thetypes of stores in this study. The stores were divided into eight categories withseparate numerical codes assigned to each, in order to serve as dummy variables. Thebroad categories were: apparel, books, computers and office supplies, hardware andhousehold goods, specialty, non-apparel, health and beauty products, and others.Table III illustrates details of the number of retailers in each category as well asexamples of web-only and multi-channel retailers in each of these categories.

    3.2 Dependent variablesWe now describe how the various dependent variables to be used in this study aregenerated in the IR500 and Bizrate.com data. The IR500 serves as the data source,except where noted.

    3.2.1 Web traffic. Web traffic is measured as the average number of unique visitorsto the online store on a monthly basis. The data are obtained from the online visitorslogs of the companies, wherever possible. Wherever data are not available from thissource, they are obtained from comScore or Neilsen/netRatings.

    Efficacy of onlinesales channels

    115

  • 3.2.2 SKUs on the web. SKU data are obtained directly from the company. In caseswhere these data are not available from the companies, they are obtained fromsecondary sources such as industry analysts and experts.

    3.2.3 Ease of finding product. The scores for ease of finding the product for which anonline shopper was looking, came from the web site Bizrate.com. BizRate.com is one ofthe most well-known price comparison web sites on the internet. It surveys onlineretailers customers and asks them to evaluate the retailers services along multipledimensions. It also searches and updates the product, price, and deal information for alarge number of online retailers daily. Data from Bizrate.com have been used quiteoften in academic research, and do not show evidence of non-response bias (Pan et al.,2002; Reibstein, 2002). On average, each of the retailers in our data set had received90,481 customer ratings using Bizrate.coms one to ten scale, with ten representingoutstanding.

    3.2.4 Conversion rate. Conversion rate is measured as the average percentage ofunique visitors to the online store who actually bought an item from the store.Most often, the IR500 collect these data directly from the retailers. In cases where thesedata are not available from the companies, they are obtained from secondary sourcessuch as industry analysts and experts.

    3.2.5 Ticket size. The ticket size of the purchase is measured as the average value ofa sale. Most often, the IR500 collect these data directly from the retailers. In caseswhere these data are not available from the companies, they are obtained fromsecondary sources such as industry analysts and experts.

    3.2.6 Outsourcing the logistics function. The extent of outsourcing the logisticsfunction was measured as the sum of four activities:

    (1) Content delivery.

    (2) Order fulfillment.

    (3) Order management.

    (4) Payment collection.

    These data are obtained directly from the retailers involved, or from industry experts.We coded each activity as 1 (outsourced) or 0 (in-house) and summed scores for the fouractivities were obtained to generate an overall score for the logistics functionas a whole. The summed score ranged from 0 (no outsourcing) to 4 (all activitiesoutsourced).

    Category No. of retailersMulti-channel example(data points)

    Web-only example(data points)

    Apparel 51 nordstrom.com (30) hats.com (21)Books 27 barnesandnoble.com (8) audible.com (19)Computers and office 36 staples.com (13) 123inkjets.com (23)Hardware 47 lowes.com (14) kitchensource.com (33)Specialty 32 samashmusic.com (13) sheetmusicplus.com (19)Non-apparel 31 sharperimage.com (21) art.com (10)Health and beauty 21 sephora.com (6) skinstore.com (15)Others/multi-specialty 15 target.com (6) igourmet.com (9)

    Table III.Product type groupings

    IJPDLM39,2

    116

  • 3.2.7 Outsourcing the target marketing function. The extent of outsourcing themarketing function is measured as the sum of four activities:

    (1) Affiliate marketing.

    (2) E-mail marketing.

    (3) E-commerce platform.

    (4) Search engine marketing.

    These data are obtained directly from the retailers involved, or from industry experts.We generated overall scores in the same fashion as they were for logistics outsourcing.

    3.2.8 Outsourcing the customer support function. Outsourcing of the customersupport function is measured as the sum of four activities:

    (1) Customer relationship management.

    (2) Web site design.

    (3) Web site search.

    (4) Web analytics.

    These data are obtained directly from the retailers involved, or from industry experts.Overall, scores were generated on a consistent basis as in logistics and marketingoutsourcing.

    4. Data analysisAfter filtering out cases for non-available or missing data and one outlier(i.e. Amazon.com), we were left with a total of 260 usable observations. Of the260 observations, 149 were web-only retailers, while 111 were multi-channel retailers.The descriptive statistics for the data are provided in Table IV, while the correlationmatrix for our dependent variables is presented in Table V.

    As can be seen, most of the correlations between the terms are quite low, thusindicating that multi-collinearity should not be an issue with the data. As a result,ordinary least squares (OLS) estimation is appropriate for this type of data. As a result,H1 to H5 were tested using a series of multivariate OLS equations, with the type ofretailer (web-only/multi-channel) serving as the independent dummy variable andeach of web traffic, SKUs on the web, ease of finding product, conversion rate, andaverage ticket serving as the dependent variables, in turn. Store size and productcategory were controls throughout all these estimations. H6 to H8 were tested using

    Multi-channel Web-onlyMean SD Mean SD

    Monthly unique visits 2,377,112 372,623 1,091,138 341,079Total SKUs on the web 226,315 92,414 191,348 89,684Ease of finding product (1-10) 8.14 0.40 8.77 0.41Conversion rate (percent) 3.55 3.69 4.96 2.61Average ticket ($) 221.36 201.73 190.64 108.28Logistics outsource (0-4) 1.51 1.28 2.05 1.3Marketing outsource (0-4) 1.53 1.38 2.48 1.66Support outsource (0-4) 1.13 1.31 1.72 1.49

    Table IV.Descriptive statistics

    Efficacy of onlinesales channels

    117

  • Mon

    thly

    un

    iqu

    ev

    isit

    s

    Tot

    alS

    KU

    son

    the

    web

    Eas

    eof

    fin

    din

    gC

    onv

    ersi

    onra

    teA

    ver

    age

    tick

    etL

    ogis

    tics

    outs

    ourc

    eM

    ark

    etin

    gou

    tsou

    rce

    Su

    pp

    ort

    outs

    ourc

    e

    Mon

    thly

    un

    iqu

    ev

    isit

    s1.

    00T

    otal

    SK

    Us

    onth

    ew

    eb2

    0.02

    1.00

    Eas

    eof

    fin

    din

    g2

    0.05

    0.03

    1.00

    Con

    ver

    sion

    rate

    0.00

    20.

    042

    0.03

    1.00

    Av

    erag

    eti

    cket

    20.

    020.

    010.

    012

    0.16

    1.00

    Log

    isti

    csou

    tsou

    rce

    0.09

    0.08

    20.

    030.

    030.

    051.

    00M

    ark

    etin

    gou

    tsou

    rce

    20.

    010.

    072

    0.12

    20.

    070.

    160.

    391.

    00S

    up

    por

    tou

    tsou

    rce

    0.01

    0.13

    20.

    022

    0.01

    0.09

    0.16

    0.46

    1.00

    Table V.Correlation matrix

    IJPDLM39,2

    118

  • the non-parametric Wilcoxon rank-sum test, since the dependent variable in thesecases was measured as sums of dichotomous scales, which thereby does not follow acontinuous scale or normal distribution.

    5. Results, implications, and future researchThe results from the data analysis are summarized in Tables VI and VII. As can be seen,seven out of the eight hypotheses were supported. Each result is now discussedindividually. H1 stated that the incoming web traffic to multi-channel retailers would behigher than that for web-only retailers. Based on the premise of TCE, we had arguedthat consumers would try to reduce their transaction costs by searching for productinformation on the internet, even when they may actually intend to purchase theproduct from the brick-and-mortar store. As the results suggest, this appears to be the casein that the hypothesis was strongly supported. Moreover, significant support forH4 seemsto further give credence to this window shopping argument; the conversion rate(percentage of visitors who actually buy) is significantly lower for multi-channel retailers,as compared to web-only retailers. We will review this observation in subsequentdiscussions.

    H2 suggested that multi-channel retailers would display more SKUs on their web site,as compared to web-only retailers, in keeping with the proposition of reducing buyersearch costs. Indeed, this hypothesis was also strongly supported. We proposed that sincemulti-channel retailers cater to a larger stream of visitors (perH1) they will have to cater toa more idiosyncratic array of tastes and preferences, consequently implying that thenumber of SKUs that they must display will be higher. Moreover, a casual browsingsession through a popular multi-channel retailer such as staples.com suggests that theyhave several products on the web site that are not available at the online store itself;however, the site directs one to the brick-and-mortar store closest to the buyer, once thebuyer decides to make a purchase. The online channel of the multi-channel retailer servesnot only as a transaction medium but also as an information medium, explaining theincreased number of SKUs on multi-channel retailers sites compared to web-only retailers.While conjecture on our part, this proposition presents an interesting avenue for futureresearch.

    Related toH2wasH3, which proposed that it would be more difficult to find particularproducts on the web storefronts of multi-channel retailers than would be for web-onlyretailers. This hypothesis was strongly supported. In keeping with the tenets of TCE, itcan thus be argued that multi-channel retailers will must provide advanced searchfunctions on their web storefronts (thereby reducing the relatively high search costs), inorder to compete effectively with web-only retailers in the online retailing space.

    Support for H4 provides further intuitive support to the notion that multi-channelretailers experience a significantly higher percentage of window shoppers thanweb-only retailers. One question that is worthy of asking is what can explain thefinding that the percentage of visitors who ultimately purchase from web-only retailersis higher than multi-channel retailers online stores? One possibility for this difference isthat the percentage of visitors to a typical multi-channel retailers online store, who visitit with the explicit intention of just looking (indicating window shopping/researchshopping behavior), is significantly higher than the corresponding percentage for aweb-only store. A competing explanation, however, is what is obtained by combiningthe results of H3 and H4; since it is relatively difficult for shoppers to find what they are

    Efficacy of onlinesales channels

    119

  • No.

    Dep

    end

    ent

    var

    iab

    leH

    yp

    oth

    esis

    bS

    ES

    td.b

    tp

    Fin

    din

    g

    H1

    Web

    traf

    fic

    Mu

    lti-

    chan

    nel.

    web

    only

    1,65

    0,14

    746

    5,30

    60.

    217

    3.54

    0.00

    0S

    up

    por

    ted

    H2

    SK

    Us

    Mu

    lti-

    chan

    nel.

    web

    only

    4,13

    3,22

    21,

    307,

    686

    0.19

    33.

    160.

    002

    Su

    pp

    orte

    dH3

    Eas

    eof

    fin

    din

    gp

    rod

    uct

    Mu

    lti-

    chan

    nel,

    web

    only

    20.

    227

    0.06

    0.22

    92

    3.77

    0.00

    0S

    up

    por

    ted

    H4

    Con

    ver

    sion

    rate

    Mu

    lti-

    chan

    nel,

    web

    only

    20.

    008

    0.00

    40.

    128

    22.

    060.

    040

    Su

    pp

    orte

    dH5

    Tic

    ket

    size

    Mu

    lti-

    chan

    nel.

    web

    only

    2.31

    423

    .68

    0.00

    60.

    090.

    922

    Not

    sup

    por

    ted

    Table VI.Results of the dataanalysis (H1 to H5)

    IJPDLM39,2

    120

  • looking for on the web storefronts of multi-channel retailers (from H3), severalprospective buyers may just give up and leave after a while. This would likely result ina lost sale, possibly to the benefit of a web-only competitor, consistent with the premiseof TCE. A direct implication then, of H3 and H4 is that multi-channel retailers will, inmany instances, need to invest more into their online retailing efforts than web-onlyretailers especially by designing highly efficient search functions, for every dollar of salegenerated especially if such research shoppers are also comparative shoppers,evaluating product offerings across retailer categories (Neslin et al., 2006).

    H5 (higher ticket size for multi-channel retailers as compared to web-only retailers) wasnot supported. While counter to our original logic, this result does offer an interesting lineof reasoning; it may be likely that the shipping-cost disparity between multi-channel andweb-only retailers studied by researchers such as Ancarani and Shankar (2004) hasreduced in magnitude over time since differences in product pricing between online andoffline channels may now not be as high as they used to be. The Ancarani and Shankar(2004) study was conducted in 2002, and it can be reasoned that the online retailingmarketplace has substantially matured in the ensuing years. For example, Circuit Cityrecently announced its One Price Promise that guarantees a uniform price for productsavailable through its various sales channels (Felberbaum, 2008). Another competingexplanation is that online customers may not notice or care about subtle differences inshipping prices, especially if a partitioned price mechanism is followed and thus may notbe interested in combining orders to bring down the per-order price (Morwitz et al., 1998).

    The final three hypotheses examined the extent of outsourcing key supportfunctions based on the premise of the retailers trying to reduce their transaction costs.H6, which stated that multi-channel retailers would outsource the logistics function toa lesser extent than web-only retailers, was strongly supported. While previousresearchers had alluded to this idea (Razzaque and Sheng, 1998; Bienstock andMentzer, 1999; Rabinovich et al., 1999), the findings were mostly prescriptive and nottested on a holistic logistics outsourcing basis using a large sample of firms andarchival data from different types of retail channels, especially within online retailenvironments. Past research (Knemeyer et al., 2003; Maltz, 1994; Stank and Maltz, 1996;Rabinovich et al., 1999) had focused on the examination of various configurations andrelations resulting from a decision of logistics outsourcing, rather than studying thedifferences in the logistics outsourcing configuration due to differences in the lastlink in the supply chain. Indeed, researchers had argued that smaller firms within anonline retail environment, may be able to improve their image in customers minds, byoutsourcing the logistics to well known firms (Esper et al., 2003). Our study suggeststhat this may, indeed, be happening in practice.

    Strong support for H6 also gives rise to a very interesting question for futureresearch; while it is likely that multi-channel retailers can piggyback on their existinginfrastructure, where is the tipping point? At what percentage of online sales, will theexisting logistics infrastructure of multi-channel retailers give way? What might such

    No. Dependent variable Hypothesis t p Finding

    H6 Logistics outsourcing Multi-channel , web only 24.154 0.000 SupportedH7 Marketing outsourcing Multi-channel , web only 23.136 0.002 SupportedH8 Support outsourcing Multi-channel , web only 22.075 0.007 Supported

    Table VII.Results of the data

    analysis (H6 to H8)

    Efficacy of onlinesales channels

    121

  • multi-channel retailers do once the percentage of their sales from online channelsexceeds that critical value? This is not a wholly hypothetical question; the online retailmarketplace has grown faster than the conventional marketplace, and the day may notbe far when online sales actually overtake conventional channel sales for many retailers.

    H7 and H8 (lower outsourcing of the marketing and customer support functionsamongst multi-channel retailers as compared to web-only retailers) were supported.We have noted that the underlying rationale for this observation could be Pisanos (1990)premise that firms with limited experience in certain areas are wise to outsource thosefunctions to specialized firms. Pisanos rationale for suggesting this argument is intuitive:firms with limited experience and knowledge in particular areas can gain, by outsourcingto specialist firms, the necessary trust in consumers minds by aligning themselves withmore specialized firms. Consequently, the policing and enforcement costs for theircustomers will be lowered, thus reducing the overall transaction costs of doing business(Goldsby and Eckert, 2003; Bowersox and Cooper, 1992). With offshoring and outsourcingof the customer support function becoming more common among American firms (Aroraand Forman, 2007), it would be reasonable to argue that web-only firms prefer to contractwith specialist vendor firms to take care of these functions. Another explanation could bethat web-only firms do not typically possess the scale economies to efficiently do themarketing and customer support functions in-house.

    Additional results with Amazon.com included in the sample are presented in theappendix. As can be seen, the hypothesis results with and without Amazon.com included inthe sample are consistent save for H2 (SKUs available), which is not supported withAmazons inclusion. The reason for lack of support for H2 is not surprising; Amazon.comhas over 1,000 percent more SKUs available on the web than the next closest competitor.Thus, arguably, the addition of Amazon.com introduces to the SKU data an outlier so largethat the overall results lose generalizeability to the immense population of internet retailers.

    6. Conclusions and limitationsIn this paper, we have studied the relative efficacies of two channel types across keymarketing and logistics variables. In addition, we have shown that several differencesexist between multi-channel and web-only retailers with respect to how their channelstructures operate. We have argued that TCE can explain several of these differencesfrom both consumers and retailers perspectives. However, one pressing issue arises asa result of the findings of this paper. While this study follows from the positivistparadigm (Hunt, 1976), it gives rise to several central normative questions. Whiledifferences appear to manifest across the two types of channel structures, how canthese differences be turned into something beneficial for retailers? For example,while we show that web-only retailers have a higher likelihood of outsourcing thelogistics function, what may happen if a particular web-only retailer chooses to keepthis function in-house and build logistics as a core competence? Can world-classlogistics become a source of competitive advantage for a web-only retailer, if otherfirms in its industry are not pursuing such an advantage? Several such interestingquestions have not been covered in this study and are left to future research. However,it should be noted that, overall, this paper has taken a first step to examine keydistinctions in performance and operational approaches across the two channel types.

    Like any other study, however, this research also has some limitations. First, it could beimplied that our sample set of firms in this study does not represent a truly random sample.

    IJPDLM39,2

    122

  • There is an obvious selection bias in the sample of firms itself. In this study, we have lookedat data from the top 500 internet retail firms, and it could be argued that our results areskewed towards larger-sized firms. We should reiterate though that our original 500 onlineretailers accounted for over 60 percent of the total online retail sales and even the reducedsample of 260 retailers accounted for over 40 percent of total online retail sales in the USA,and thus, we believe that the data are fairly representative of the US market as a whole, ifnot of the world market. At the same time, we also agree that future research shouldnevertheless try to overcome this limitation and try to see if our findings hold with truly arandom sample of online retailers.

    A second limitation of the study is the nature of the dataset itself. The data in severalcases consist of self-report numbers provided by the firms themselves (e.g. sales data).While every care has been maintained to retain the integrity of the data by dropping thosecases which were obtained based purely on expert opinions, there is, at the moment, nomethod by which we can ensure that the self-reported numbers given by firms are indeedaccurate. Though a weakness of this study, this limitation exists in many forms ofbusiness research. For example, in survey research, it is difficult to ensure that the surveyis answered by the person to whom it is addressed, or that the person has answeredknowledgeably and truthfully. Future research should, nevertheless, address thislimitation of our data.

    A third key limitation of our data is the way the indices for outsourcing have beenoperationalized. In this study, there could only be two options for any individualactivity; either it was fully outsourced (coded as 1) or it was fully conducted in-house(coded as 0). Subsequently, the scores on each of the individual variables within theprocess functions (logistics, marketing, customer support) were summed to arrive at acomposite score, which ranged between 0 and 4. Arguably, this method of measuringoutsourcing is quite simplistic; often times, an activity is neither fully outsourced norperformed entirely in-house, it is partially outsourced. For example, some firms mayhire specialized vendors for activities like packaging operations during periods of highdemand (e.g. Christmas/New Year), and at other normal times, prefer to do the activitywith its own staff. Similarly, some firms may choose to deliver goods to theirhigh-value customers themselves, and outsource delivery to lower-value customers.We have not measured this type of partial outsourcing in this study and futureresearch should measure this variable when investigating this topic.

    Despite these limitations and shortcomings, this study has taken an important stepforward in studying the differences in supply chain operations between web-only andmulti-channel retailers. We believe that this study and its line of inquiry were overdue;it has been over a decade since the earliest online retailers such as Amazon.com andoverstock.com made their debuts in the marketplace. However, insofar as comparingoperations between web-only and multi-channel retailers is concerned, research haslagged behind. This study has taken an important step in bridging this gap inknowledge. For example, researchers have argued that analyses [need to be]performed on various e-tailing processes, which may help better identify which typesof e-tailers are efficient on what processes (e.g. procurement, fulfillment,communication of value proposition) [in order to] better understand the mechanisms(and processes) that need to be followed in order to be successful (Grewal et al., 2004).This paper is a first step towards that objective.

    Efficacy of onlinesales channels

    123

  • Notes

    1. For an interesting study in this context, refer to Slack et al. (2008). For other research on suchinformation seeking behavior (Van Del Poel and Leunis, 1999).

    2. A note on the need for controls: product type is an important criterion in channel choice(Schoenbachler and Gordon, 2002; Sullivan and Thomas, 2008). Some customers may preferto buy certain types of products in a brick-and-mortar environment so that they can actuallyexperience the products, assess their quality and make easy returns (Young, 2001; Peck andChilders, 2003; Laseter et al., 2006). Thus, it is important to control for product type, as thesemay directly impact channel dynamics (Laseter et al., 2006). Similarly, it could be argued thatlarger sized stores would receive more traffic, owing to better internet visibility throughsearch engines. Thus, it becomes important to control for store size as well.

    References

    Alba, J., Lynch, J.G., Weitz, B., Janiszewski, C., Lutz, R. and Wood, S. (1997), Interactive homeshopping: consumer, retailer, and manufacturer incentives to participate in electronicmarketplaces, Journal of Marketing, Vol. 61 No. 3, pp. 38-53.

    Alptekinoglu, A. and Tang, C. (2005), A model for analyzing multi-channel distributionsystems, European Journal of Operational Research, Vol. 163, pp. 802-24.

    Ancarani, F. and Shankar, V. (2004), Price levels and price dispersion within and across multipleretailer types: further evidence and extension, Journal of the Academy of MarketingScience, Vol. 32 No. 2, pp. 176-88.

    Ang, S. and Straub, D. (1998), Production and transaction economies and IS outsourcing: a studyof the US banking industry, MIS Quarterly, Vol. 22 No. 4, pp. 535-52.

    Ansari, A., Mela, C. and Neslin, S. (2008), Customer channel migration, Journal of MarketingResearch, Vol. 45 No. 1, pp. 60-76.

    Arora, A. and Forman, C. (2007), Proximity and information technology outsourcing how localare IT markets, Journal of Management Information Systems, Vol. 24 No. 2, pp. 73-102.

    Aubert, B., Rivard, S. and Patry, M. (2004), A transaction cost model of IT outsourcing,Information & Management, Vol. 41 No. 7, pp. 921-32.

    Bakos, Y. (1997), Reducing buyer search costs: implications for electronic marketplaces,Management Science, Vol. 43 No. 12, pp. 1676-92.

    Bannon, L. (2000), Selling Barbie online may pit Mattel vs. stores, The Wall Street Journal,November 17, p. B1.

    Bettman, J., Luce, M.F. and Payne, J. (1998), Constructive consumer choice processes, Journal ofConsumer Research, Vol. 25 No. 3, pp. 187-217.

    Bienstock, C.C. and Mentzer, J.T. (1999), An experimental investigation of the outsourcingdecision for motor carrier transportation, Transportation Journal, Vol. 39 No. 1, pp. 42-59.

    Bolumole, Y., Frankel, R. and Dag, N. (2007), Developing a theoretical framework for logisticsoutsourcing, Transportation Journal, Vol. 46 No. 2, pp. 35-54.

    Bowersox, D.J. and Cooper, M.B. (1992), Strategic Marketing Channels, McGraw-Hill,Burr Ridge, IL.

    Boyer, K. and Hult, T. (2005), Consumer behavior in an online ordering application: a decisionscoring model, Decision Sciences, Vol. 36 No. 4, pp. 569-98.

    Brohan, J. (2008), The top 500 guide, June, 2008, available at: www.internetretailer.com/article.asp?id 22579 (accessed July 30).

    IJPDLM39,2

    124

  • Brynjolfsson, E. and Smith, M.D. (2000), Frictionless commerce? A comparison of internet andconventional retailers, Management Science, Vol. 46 No. 4, pp. 563-85.

    Chircu, A. and Mahajan, V. (2006), Managing electronic commerce retail transaction costs forcustomer value, Decision Support Systems, Vol. 42 No. 2, pp. 898-914.

    Choudhury, V. and Karahanna, E. (2008), The relative advantage of electronic commerce:a multidimensional view, MIS Quarterly, Vol. 22 No. 1, pp. 471-508.

    Choudhury, V., Hartzel, K. and Konsynski, B. (1998), Uses and consequences of electronicmarkets: an empirical investigation in the aircraft parts industry, MIS Quarterly, Vol. 22No. 4, pp. 471-508.

    Clark, B.H. (1997), Welcome to my parlor: the lure of marketing on the world wide web is great.Be sure you dont get stuck with wrong approach, Marketing Management, Vol. 5 No. 4,pp. 11-22.

    Clark, T.H.K. and Kauffman, R.J. (2000), Introduction to the special section: electronicintermediaries and networks in business-to-business electronic commerce, InternationalJournal of Electronic Commerce, Vol. 4 No. 4, pp. 5-6.

    Cleary, M. (2000), The promise of multichannel retailing, Inter@ctive Week, Vol. 7, p. 50.

    Coase, R. (1937), The nature of the firm, Economica, Vol. 4 No. 16, pp. 386-405.

    DMA (2005), 2005Multichannel Marketing Report, Direct Marketing Association, New York, NY.

    Devaraj, S., Fan, M. and Kohli, R. (2006), Examination of online channel preference: using thestructure-conduct-outcome framework, Decision Support Systems, Vol. 42 No. 2,pp. 1089-103.

    Esper, T., Jensen, T., Turnipseed, F. and Burton, S. (2003), The last mile: an examination ofeffects of online retail delivery strategies on consumers, Journal of Business Logistics,Vol. 24 No. 2, pp. 177-203.

    Felberbaum, M. (2008), Circuit city to offer the same price in stores, on web, The Boston Globe,October 14, available at: www.boston.com/business/articles/2008/10/14/circuit_city_to_offer_same_price_in_stores_on_web/ (accessed October 15).

    Forster, S. (2004), When one hand doesnt. . ., The Wall Street Journal, March 22, p. R3.

    Gabrielsson, M., Kirpalani, V.H.M. and Luostarinen, R. (2002), Multiple channel strategies in theEuropean personal computer industry, Journal of International Marketing, Vol. 10 No. 3,pp. 73-95.

    Gensler, S., Dekimpe, M.G. and Skiera, B. (2004), Evaluating channel performancein multi-channel environments, Journal of Retailing and Consumer Services, Vol. 14No. 1, pp. 17-23.

    Goldsby, T.J. and Eckert, J.A. (2003), Electronic transportation marketplaces: a transaction costperspective, Industrial Marketing Management, Vol. 32 No. 3, pp. 187-98.

    Grewal, D., Iyer, G. and Levy, M. (2004), Internet retailing: enablers, limiters, and marketingconsequences, Journal of Business Research, Vol. 57 No. 7, pp. 703-13.

    Grover, V., Ramanlal, P. and Segars, A.H. (1999), Information exchange in electronic markets:implications for market structures, International Journal of Electronic Commerce, Vol. 3No. 4, pp. 89-102.

    Hobbs, J.E. (1996), A transaction cost approach to supply chain management, Supply ChainManagement, Vol. 1 No. 2, pp. 15-27.

    Howlett, G. (2008), Internet retailer top 500 released, Marketing Pilgrim, available at: www.marketingpilgrim.com/2008/06/internet-retailer-top-500-guide.html (accessed July 30, 2008).

    Efficacy of onlinesales channels

    125

  • Hunt, S. (1976), The nature and scope of marketing, Journal of Marketing, Vol. 40 No. 3,pp. 17-28.

    Ivanaj, V. and Franzil, Y.M. (2006), Outsourcing logistics activities: a transaction cost economicsperspective, Proceedings of the 15th International Conference of Strategic Management,Geneva, June 13-16.

    Janssen, M. and Sol, H.G. (2000), Evaluating the role of intermediaries in the electronic valuechain, Internet Research, Vol. 10 No. 5, pp. 406-17.

    Keeney, R.L. (1999), The value of internet commerce to the customer, Management Science,Vol. 45 No. 4, pp. 533-42.

    Kim, J. and Park, J. (2005), A consumer shopping channel extension model: attitude shift towardthe online store, Journal of Fashion Marketing & Management, Vol. 9 No. 1, pp. 106-21.

    Klein, L. and Ford, G. (2003), Consumer search for information in the digital age: an empiricalstudy of prepurchase search for automobiles, Journal of Interactive Marketing, Vol. 17No. 3, pp. 29-49.

    Klein, S., Frazier, G.L. and Roth, V.J. (1990), A transaction cost analysis model of channelintegration in international markets, Journal of Marketing Research, Vol. 27 No. 2,pp. 196-208.

    Knemeyer, A.M., Corsi, T.M. and Murphy, P.R. (2003), Logistics outsourcing relationships:customer perspectives, Journal of Business Logistics, Vol. 24 No. 1, pp. 77-109.

    Kohli, R., Devaraj, S. and Mahmood, M.A. (2004), Understanding determinants of onlineconsumer satisfaction: a decision process perspective, Journal of ManagementInformation Systems, Vol. 21 No. 1, pp. 115-35.

    Kumar, N. and Ruan, R. (2006), On manufacturers complementing the traditional retail channelwith a direct online channel, Quantitative Marketing and Economics, Vol. 4 No. 3,pp. 289-323.

    Kumar, V. and Venkatesan, R. (2005), Who are multichannel shoppers and how do theyperform? Correlates of multichannel shopping behavior, Journal of Interactive Marketing,Vol. 19 No. 2, pp. 44-61.

    Laseter, T.M., Rabinovich, E. and Huang, A. (2006), The hidden costs of clicks, Strategy Business, Vol. 42, pp. 26-30.

    Laseter, T.M., Rabinovich, E., Boyer, K.K. and Rungtusanatham, M. (2007), The future of theweb: three critical issues in internet retailing, MIT Sloan Management Review, Vol. 48No. 3, pp. 58-64.

    Lee, B.K. and Lee, W.N. (2004), The effect of information overload on consumer choice qualityin an on-line environment, Psychology & Marketing, Vol. 21 No. 3, pp. 159-84.

    Liang, T. and Huang, J. (1998), An empirical study on consumer acceptance of productsin electronic markets: a transaction cost model, Decision Support Systems, Vol. 24 No. 1,pp. 29-43.

    Lynch, J.G. and Ariely, D. (2000), Wine online, Marketing Science, Vol. 19 No. 1, pp. 83-103.

    Malone, T., Yates, J. and Benjamin, R. (1987), Electronic markets and electronic hierarchies,Communications of the ACM, Vol. 30 No. 6, pp. 484-97.

    Maltz, A.B. (1993), Private fleet use: a transaction cost model, Transportation Journal, Vol. 33No. 2, pp. 12-19.

    Maltz, A.B. (1994), The relative importance of cost and quality in the outsourcingof warehousing, Journal of Business Logistics, Vol. 15 No. 2, pp. 45-62.

    IJPDLM39,2

    126

  • Montoya-Weiss, M.M., Voss, G.B. and Grewal, D. (2003), Determinants of online channel useand overall satisfaction with a relational, multichannel service provider, Journal of theAcademy of Marketing Science, Vol. 31 No. 4, pp. 448-58.

    Morwitz, V.G., Greenleaf, E.A. and Johnson, E.J. (1998), Divide and prosper: consumersreactions to partitioned prices, Journal of Marketing Research, Vol. 35 No. 4, pp. 453-63.

    Muthitacharoen, A., Gillenson, M. and Suwan, N. (2006), Segmenting online customersto manage business resources: a study of the impacts of sales channel strategieson consumer preferences, Information & Management, Vol. 43 No. 5, pp. 678-95.

    Neslin, S., Grewal, D., Leghorn, R., Shankar, V., Teerling, M., Thomas, J. and Verhoef, P.C. (2006),Challenges and opportunities in multichannel customer management, Journal of ServiceResearch, Vol. 9 No. 2, pp. 95-112.

    Noble, S., Griffith, D. and Weinberger, M. (2005), Consumer derived utilitarian value andchannel utilization in a multi-channel retail context, Journal of Business Research, Vol. 58No. 12, pp. 1643-51.

    Pache, G. and Sauvage, T. (1999), Logistics: Strategic Issues, 2nd ed., Vuibert Entreprises, Paris.

    Pan, X., Ratchford, B. and Shankar, V. (2002), Can price dispersion in online markets beexplained by differences in e-tailer service quality?, Journal of the Academy of MarketingScience, Vol. 30 No. 4, pp. 433-45.

    Peck, J. and Childers, T.L. (2003), To have and to hold: the influence of haptic informationon product judgments, Journal of Marketing, Vol. 67 No. 2, pp. 35-48.

    Pisano, G.P. (1990), The R&D boundaries of the firm: an empirical analysis, AdministrativeScience Quarterly, Vol. 35 No. 1, pp. 153-76.

    Rabinovich, E. and Evers, P. (2003), Product fulfillment in supply chains supportinginternet-retailing operations, Journal of Business Logistics, Vol. 24 No. 2, pp. 205-36.

    Rabinovich, E., Bailey, J.P. and Carter, C. (2003), A transaction-efficiency analysis of an internetretailing supply chain in the music CD industry, Decision Sciences, Vol. 34 No. 1,pp. 131-72.

    Rabinovich, E., Windle, R., Dresner, M. and Corsi, T. (1999), Outsourcing of integrated logisticsfunctions: an examination of industry practices, International Journal of PhysicalDistribution & Logistics Management, Vol. 29 No. 6, pp. 353-73.

    Ratchford, B., Lee, M. and Talukdar, D. (2003), The impact of the internet on information searchfor automobiles, Journal of Marketing Research, Vol. 40 No. 2, pp. 193-209.

    Razzaque, M.A. and Sheng, C.C. (1998), Outsourcing of logistics functions: a literature survey,International Journal of Physical Distribution & Logistics Management, Vol. 28 No. 2,pp. 89-107.

    Reibstein, D. (2002), What attracts customers to online stores and what keeps them comingback?, Journal of the Academy of Marketing Science, Vol. 30 No. 4, pp. 465-73.

    Sarkar, M., Butler, B. and Steinfield, C. (1998), Cybermediaries in the electronic marketspace:toward theory building, Journal of Business Research, Vol. 41 No. 3, pp. 215-21.

    Saunders, C. (2002), Survey: multi-channel buyers worth pursuing, ClickZ, January 28,p. 30, available online at: www.clickz.com/963321 (accessed July 30, 2008).

    Schoenbachler, D.D. and Gordon, G.L. (2002), Multi-channel shopping: understanding whatdrives channel choice, Journal of Consumer Marketing, Vol. 19 No. 1, pp. 42-53.

    Sharma, A. and Sheth, J. (2004), Web-based marketing: the coming revolution in marketingthought and strategy, Journal of Business Research, Vol. 57 No. 7, pp. 696-702.

    Efficacy of onlinesales channels

    127

  • Simon, H.A. (1990), Invariants of human behavior, Annual Review of Psychology, Vol. 41,pp. 1-19.

    Slack, F., Rowley, J. and Coles, S. (2008), Consumer behavior in multi-channel contexts: the caseof a film festival, Internet Research, Vol. 18 No. 1, pp. 46-59.

    Srinivasan, S., Anderson, R. and Ponnavolu, K. (2002), Consumer loyalty in e-commerce:an exploration of its antecedents and consequences, Journal of Retailing, Vol. 78 No. 1,pp. 41-50.

    Stank, T.P. and Maltz, A.B. (1996), Some propositions on third party choice: domestic v/sinternational logistics providers, Journal of Marketing Theory and Practice, Vol. 4 No. 2,pp. 45-54.

    Stern, L.W., El-Ansary, A.I. and Coughlan, A.T. (1996), Marketing Channels, 5th ed.,Prentice-Hall, Upper Saddle River, NJ.

    Sullivan, U. and Thomas, J. (2008), Customer migration: an empirical investigation acrossmultiple channels, working paper, University of Illinois (UIUC), Urbana-Champaign, IL,available at: www.business.uiuc.edu/Working_Papers/papers/04-0112.pdf

    Swaminathan, V., Lepkowska, W., Elzbietam, R. and Bharat, P. (1999), Browsers or buyers incyberspace? An investigation of factors influencing electronic exchange, Journal ofComputer Mediated Communication, Vol. 5 No. 2, available at: www.ascusc.org/jcmc/vol5/issue2/oldswaminathan.html

    Swinyard, W. and Smith, S.M. (2003), Why people (dont) shop online: a lifestyle study of theinternet consumer, Psychology & Marketing, Vol. 20 No. 7, pp. 597-657.

    Tang, F.F. and Xing, X. (2001), Will the growth of multi-channel retailing diminish the pricingefficiency of the web?, Journal of Retailing, Vol. 77 No. 3, pp. 319-33.

    Tsay, A. and Agarwal, N. (2004), Channel conflict and coordination in the e-commerce age,Production and Operations Management, Vol. 13 No. 1, pp. 93-110.

    Van den Poel, D. and Leunis, J. (1999), Consumer acceptance of the internet as a channel ofdistribution, Journal of Business Research, Vol. 45 No. 3, pp. 249-56.

    Verhoef, P.C., Neslin, S.A. and Vroomen, B. (2007), Browsing versus buying: determinants ofcustomer search and buy decisions in a multi-channel environment, working paper,University of Groningen, Groningen.

    Williamson, O.E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications,Free Press, New York, NY.

    Williamson, O.E. (1979), Transaction-cost economics: the governance of contractual relations,Journal of Law & Economics, Vol. 22 No. 2, pp. 233-61.

    Williamson, O.E. (1985), Economic Institutions of Capitalism: Firms, Markets, and RelationalContracting, The Free Press, New York, NY.

    Wu, D., Ray, G. and Whinston, A. (2008), Manufacturers distribution strategy in the presence ofan electronic channel, Journal of Management Information Systems, Vol. 25 No. 1,pp. 167-98.

    Xing, P., Ratchford, B.T. and Shankar, V. (2004), Price dispersion on the internet: a review anddirections for future research, Journal of Interactive Marketing, Vol. 18 No. 4, pp. 116-35.

    Young, S.J. (2001), Make the most of every channel, Catalog Age, Vol. 18, pp. 58-9.

    Zeithaml, V., Parasuraman, A. and Malhotra, A. (2002), Service quality delivery throughweb sites: a critical review of extant knowledge, Journal of the Academy of MarketingScience, Vol. 30 No. 4, pp. 362-75.

    IJPDLM39,2

    128

  • Appendix. Test results with Amazon.com included in the sample

    No.

    Dep

    end

    ent

    var

    iab

    leH

    yp

    oth

    esis

    bS

    ES

    td.b

    tp

    Fin

    din

    g

    H1

    Web

    traf

    fic

    Mu

    lti-

    chan

    nel.

    web

    only

    1,34

    8,19

    959

    7,53

    30.

    140

    2.25

    60.

    025

    Su

    pp

    orte

    dH2

    SK

    Us

    Mu

    lti-

    chan

    nel.

    web

    only

    2,99

    5,15

    81,

    925,

    669

    0.09

    61.

    555

    0.12

    1N

    otsu

    pp

    orte

    dH3

    Eas

    eof

    fin

    din

    gp

    rod

    uct

    Mu

    lti-

    chan

    nel,

    web

    only

    20.

    228

    0.06

    20.

    230

    23.

    804

    0.00

    0S

    up

    por

    ted

    H4

    Con

    ver

    sion

    rate

    Mu

    lti-

    chan

    nel,

    web

    only

    20.

    008

    0.00

    40.

    128

    22.

    060

    0.04

    0S

    up

    por

    ted

    H5

    Tic

    ket

    size

    Mu

    lti-

    chan

    nel.

    web

    only

    2.56

    923

    .605

    0.00

    70.

    109

    0.91

    3N

    otsu

    pp

    orte

    d

    Table AI.Regression results

    (H1 to H5)

    Efficacy of onlinesales channels

    129

  • About the authorsShashank Rao is a PhD candidate in the Department of Decision Science and InformationSystems at the University of Kentucky. His research interests lie in the areas of Logistics, SupplyChain Management, and Online Retailing. His research has appeared or is forthcoming injournals such as the International Journal of Logistics Management, Journal of ElectronicCommerce Research, and The Information Society. He has also regularly presented and publishedhis research at the CSCMP, DSI, and POMS conferences. Shashank Rao is the correspondingauthor and can be contacted at: [email protected]

    Thomas J. Goldsby is Associate Professor of Supply Chain Management at the University ofKentucky. His research interests focus on logistics customer service, supply chain integration,and the theory and practice of lean and agile supply chain strategies. He has published severalarticles in academic and professional journals and serves as a frequent speaker at academicconferences, executive education seminars, and professional meetings. He is co-author of LeanSix Sigma Logistics: Strategic Development to Operational Success (J. Ross Publishing, 2005) anda research associate of the Global Supply Chain Forum at The Ohio State University. He hasreceived recognitions for excellence in teaching at Iowa State University, The Ohio StateUniversity, and the University of Kentucky.

    Deepak Iyengar is Assistant Professor of Supply Chain Management at the University ofKentucky. He holds a PhD in Supply Chain Management from the University of Maryland,College Park. His research interests lie in the areas of logistics, distribution channelsand operations management. He also has extensive research and consulting experience inlean and six sigma methodologies.

    No.Dependentvariable Hypothesis t p Finding

    H6 Logisticsoutsourcing

    Multi-channel , web only 24.211 0.000 Supported

    H7 Marketingoutsourcing

    Multi-channel , web only 23.224 0.001 Supported

    H8 Supportoutsourcing

    Multi-channel , web only 22.790 0.005 Supported

    Note: Wilcoxon rank-sum results (H6 to H8)Table AII.

    IJPDLM39,2

    130

    To purchase reprints of this article please e-mail: [email protected] visit our web site for further details: www.emeraldinsight.com/reprints