POMSPRODUCTION AND OPERATIONS MANAGEMENTVol. 15, No. 1, Spring 2006, pp. 40 –56issn 1059-1478 兩 06 兩 1501 兩 040 1.25 2006 Production and Operations Management SocietyBoiling Frogs: Pricing Strategies for aManufacturer Adding a Direct Channel thatCompetes with the Traditional ChannelKyle Cattani Wendell Gilland Hans Sebastian Heese Jayashankar SwaminathanThe Kelley School of Business, Indiana University, Bloomington, Indiana 47405-1701, USAThe Kenan-Flagler Business School, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina27599-3490, USAThe Kelley School of Business, Indiana University, Bloomington, Indiana 47405-1701, USAThe Kenan-Flagler Business School, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina27599-3490, USAWe analyze a scenario where a manufacturer with a traditional channel partner opens up a directchannel in competition with the traditional channel. We first consider that in order to mitigatechannel conflict the manufacturer, who chooses wholesale prices as a Stackelberg leader, commits tosetting a direct channel retail price that matches the retailer’s price in the traditional channel. We findthat the specific equal-pricing strategy that optimizes profits for the manufacturer is also preferred by theretailer and customers over other equal-pricing strategies. We next consider the implications of theequal-pricing constraint through a numerical experiment that indicates that the equal-pricing strategy isappropriate as long as the Internet channel is significantly less convenient than the traditional channel.If the Internet channel is of comparable convenience to the traditional channel, then the manufacturerhas tremendous incentive to abandon the equal-pricing policy, at great peril to the traditional retailer.Key words: supply chain management; direct channel; channel conflict; pricingSubmissions and Acceptance: Submitted January 2004; revised September 2004, November 2004, January2005; accepted February 2005 by Eric Johnson.1.Introductionmanufacturers offer the product for sale over the Internet, but only at full-retail price, presumably to prevent channel conflict with existing distributors. Forexample, a replacement tricolor ink cartridge for thewidely used Hewlett Packard (HP) Deskjet 600 seriesprinter can be obtained for 31.99 at the manufacturer’s (HP) web site. This price matches exactly thenon-sale price at HP’s traditional retailers such as BestBuy.In Table 1, we display a sample of recent pricescharged for several products that are sold boththrough a manufacturer’s web site and through traditional retailers. For these examples of products in several categories, including electronics, athletic equipment, and toys, we note that the Internet price chargedby the manufacturer and the in-store price charged bya traditional retailer are exactly or nearly identical. WeThe introduction by a manufacturer of a new, Internetbased distribution channel provides an opportunity toincrease sales and/or decrease costs. An increasedcustomer base may become accessible, and the marginal cost of reaching these customers may be negligible. On the other hand, the introduction of a newchannel may threaten existing channel relationships.The challenge of appeasing current channel partners isproblematic for many manufacturers that might otherwise quickly adapt to and experiment with an Internet-based outlet. Some manufacturers, such as Daimler-Chrysler, Nikon, and Rubbermaid, thus have usedthe Internet as a medium to provide information abouttheir products and/or to point the Internet surfer tothe nearest retailer carrying the product, but do notoffer the product for sale over the Internet. Many other40

Cattani et al.: Boiling Frogs: Pricing Strategies for a Manufacturer Adding a Direct Channel that Competes with the Traditional ChannelProduction and Operations Management 15(1), pp. 40 –56, 2006 Production and Operations Management SocietyTable 1Price Comparison of Selected Products (September, 2003)ProductSony 32 Flat Tube HDTVSony 57 HD Projection TVAdidas Climacool 2 M ShoesNike Air Kantara ShoesNike Triax Stamina WatchLEGO Snowboard SuperpipeManufacturer’sweb priceTraditionalretail price 1,299.99 2,299.99 100.00 120.00 129.00 49.99 1,299.991 2,299.991 99.992 119.992 129.002 49.2331Best Buy.Dick’s Sporting Goods.3Wal-Mart.2are unaware of any comprehensive empirical studycomparing prices between manufacturer-owned websites and traditional retail locations, although a recentsurvey by Ernst and Young (2001) reports that nearlytwo thirds of companies price products identically fortheir on-line and off-line operations.In some product categories, we are beginning to seeexamples where prices on the manufacturer’s websiteare lower than prices in the traditional retail channel.The CD Los Lonely Boys can be purchased at Sony.comfor 10.98, while it was simultaneously priced at 13.88 at Wal-Mart and 13.99 at Best Buy. QuickenDeluxe, a personal financial software program, isavailable online at for 49.95, yet costs 10 more at both Best Buy and Office Depot. As will bediscussed later, our research provides theoretical justification for why this pricing behavior may be emanating from the music and software industries.This paper is motivated from our interactions withmanagers at a leading computer manufacturer whowere considering the introduction of a direct channelthat would be in competition with their traditionalchannel. The managers sought insights about the effect of different pricing strategies on profitability toboth the manufacturer and the traditional channelpartners.We thus explore a scenario where a manufacturerwith a traditional channel partner (i.e., a retailer)opens a direct channel that is in competition with thetraditional channel. For our discussion, we will consider the direct channel to be Internet based, althoughit could be any direct channel. We begin with theassumption that the manufacturer, who is a Stackelberg leader, chooses wholesale prices for the traditional channel along with retail prices for the directchannel in such a way that the retail prices in thetraditional channel and the direct channel are identical, consistent with prevalent practice.Given this general assumption of equal prices acrossthe two channels, we use consumer utility theory todevelop a model that determines the effect on profitsand market shares of the manufacturer and the retailer41for specific strategies that keep wholesale prices asthey were before (Strategy 1), keep retail prices as theywere before (Strategy 2), or set wholesale and retailprices to optimize the manufacturer’s profit (Strategy3). For each of these strategies, we determine how theresulting prices compare to the single-channel prices,and we determine the resulting profits for the supplychain and its two players. Counter to our intuition, wefind that the pricing strategy that optimizes the manufacturer’s profit (Strategy 3) also tends to be preferred by the retailer over the other two strategies.Surprisingly, the end customer also prefers Strategy 3in most cases, benefiting from lower prices. This is incontrast to Strategy 1, which creates upward pressureon retail prices. In general, we conclude that under anequal-pricing framework, the retailer does not need toview the manufacturer’s addition of a direct channelas harmful competition but rather as a mechanism forsegmenting the market in a way that benefits both themanufacturer and the retailer.We next show, through a numerical experiment,that the equal-pricing strategy is reasonable for themanufacturer as long as the Internet channel is significantly less convenient and/or more costly than theretail channel. If the Internet channel becomes moreconvenient and less costly than the traditional channel, we show that an equal-pricing strategy significantly restricts profits for the manufacturer, who atthis point has tremendous incentive to use the directchannel to undercut traditional channel prices. In doing so, the manufacturer’s large gains in profit come atthe retailer’s expense. Thus, the manufacturer’s introduction of a direct channel that competes with a traditional channel has very different effects on the retailpartner depending on the manufacturer’s pricingstrategy, which itself is affected by the relative costsand convenience of the two channels.We propose an analogy to the parable of the boilingfrog. The parable states that a frog thrown into a pot ofboiling water will quickly jump out. But a frog throwninto a pot of temperate water may stay even if thetemperature is slowly raised to boiling, leading to theuntimely demise of the frog. By introducing an Internet channel with equal pricing, the manufacturer hasplaced the retailer in a mildly competitive positionwhere the retailer may even benefit if the Internet ismore costly and less convenient on average to thepopulation of customers. But if, as we imagine, thecosts and average convenience of the Internet channelbecome more favorable over time, then the manufacturer will be in a position to use the direct channel toundercut the prices in the traditional channel, and“boil” the traditional retailer.The rest of the paper is organized as follows. InSection 2, we discuss related literature. We introduce

42Cattani et al.: Boiling Frogs: Pricing Strategies for a Manufacturer Adding a Direct Channel that Competes with the Traditional ChannelProduction and Operations Management 15(1), pp. 40 –56, 2006 Production and Operations Management Societythe demand generation model and present our analytical model and results in Section 3. We provide computational insights in Section 4, and conclude in Section 5.2.Related LiteratureThere are a number of recent papers that study issuesrelated to supply chains in e-business. Swaminathanand Tayur (2003) in their comprehensive review ofanalytical models related to Internet-based supplychains identify managing multiple distribution channels (that include an Internet channel) as one of thekey areas of supply chain research. Cattani et al.(2004b) survey recent research related to the coordination of traditional and Internet supply chains whileTsay and Agrawal (2004a) review competitive modelsof traditional and Internet supply chains.Jeuland and Shugan (1983) define channel coordination as the setting of all manufacturer- and retailercontrolled variables at the levels that maximize channel profits. Our paper relates to work done towardunderstanding the conflicts that arise given the different objectives of channel members. In contrast tomuch of the previous research, our paper assumes thatthe manufacturer is not only a channel partner of theretailer, but also competes with the retailer for endcustomer sales.In this paper, we extend the model introduced byCattani et al. (2004a), who study a scenario where aretailer offers identical products via two channels thatare perceived as different by the customers (e.g., traditional and Internet). In their model, customer utilityfor a specific product is decreasing in the price of theproduct and in the effort required (based on the particular channel) to purchase the product. Customersare not excluded from a specific channel— both channels have an opportunity to compete for all customers.Using a view that a product is a compilation of product and delivery attributes, the market is segmentedbased on the utility each customer attains from products delivered through either the direct channel or thetraditional channel. In this current work, we applytheir model in a different setting where a manufacturer with a traditional retail channel partner introduces a second channel in competition to the traditional channel.Tsay and Agrawal (2004b) consider the channel conflicts that arise as a company engages in direct sales incompetition with existing reseller partners. Theirmodel allows demand in both channels to be stimulated by costly sales effort in either channel, with thedirect channel facing an inherent cost disadvantagefor providing sales effort. They assume the retail priceis given, and focus attention on the impact of channelstructure on optimal sales effort and the resultingprofits. They conclude that the addition of a directchannel alongside a reseller channel is not necessarilydetrimental to the reseller, particularly when the direct channel faces a significant disadvantage in thecost of sales effort and/or advantage in cost of fulfillment. Our model explicitly captures the heterogeneityin customer preferences for each channel, which Tsayand Agrawal acknowledge, but do not incorporateinto their model. We also allow the retail price tochange, which enables the supply chain participants tomore fully react to the altered channel structure. Whenthe direct channel is relatively inconvenient, our results are similar to Tsay and Agrawal in that we findthat the addition of a direct channel is not necessarilydetrimental to the reseller. But if the direct channel isrelatively convenient, we find that the manufacturerhas incentive to set wholesale and retail prices thatadversely affect reseller profits.Price competition between independent retailersand manufacturer-owned stores is also studied byAhn et al. (2002). Their model assumes that manufacturer-owned stores are geographically isolated fromthe traditional retailers (i.e., they consider outlet storeson the periphery of metropolitan areas). The equilibrium pricing strategy is always one of three: monopoly, competition, or elimination. Our research followsup on the idea of market segmentation with a somewhat different approach. In our model, customer’s netutility depends on the innate utility of the productreduced by the price and the effort required to acquirethe product. In a traditional channel, effort might beassociated with the physical distance from the traditional retailers, while in a direct channel effort mightbe associated with ease, convenience, and customercomfort with the channel. The market is segmentedaccording to the utility each customer attains fromeither the direct channel or the traditional channel. Inour model, customers are not excluded from a specificchannel; thus both channels compete for all customers.Bell et al. (2003) explore the effect of partial integration on retailer behavior. Using a linear demandmodel that considers prices and efforts of independentretailers competing against an integrated retailer (whoinvests in brand support), they find that pricescharged by the independent retailer are higher thanthe prices charged when competition is between independent retailers only. While we do not considerbrand support, we also find that retail prices can behigher with the entry of an integrated channel, but inour utility driven demand model this scenario ariseswhen the manufacturer acts suboptimally.Boyaci (2005) considers a setting similar to ours,where a manufacturer sells through both a directchannel and a traditional retail channel, but his research focuses on the misaligned stocking decisionsunder demand uncertainty and on developing mech-

Cattani et al.: Boiling Frogs: Pricing Strategies for a Manufacturer Adding a Direct Channel that Competes with the Traditional ChannelProduction and Operations Management 15(1), pp. 40 –56, 2006 Production and Operations Management Societyanisms for supply chain coordination. In another paper with a focus on inventory management, Seifert etal. (2004) assume that a manufacturer has a direct(Internet) market that serves a different customer segment than the traditional retail channel (not owned bythe manufacturer). They model the inventory decisions for a supply chain comprised of a manufacturerand N retailers. They compare a dedicated supplychain to a cooperative supply chain. In a dedicatedsupply chain, the inventory sent to each retailer isused exclusively to meet demand at that retailer whileinventory at the Internet store is used exclusively tomeet demand at that store. In a cooperative supplychain, leftover inventory at the retailers is available tofill demand at the Internet store. They develop modelsthat allow them to compare the benefits of supplychain integration versus the benefits of supply-chaincooperation. In our models, the manufacturer and theretailer compete in the same market.Chiang et al. (2003) conceptualize the impact onsupply chain design of customer acceptance of a directchannel as a substitute for shopping at a traditionalstore. They conclude that direct marketing, which indirectly increases the flow of profits through the retailchannel, helps the manufacturer improve overall profitability by reducing the degree of double marginalization. The direct channel may not always be detrimental to the retailer since it will be accompanied bya wholesale price reduction. We model an identicalsupply chain structure, but use an alternate model thatallows the direct channel to be preferred to the traditional channel, for at least some customers. Our modelalso differs in how we model the customer’s willingness to pay in each channel: we model the customerutility for the two channels as separate (and independent) dimensions, while they consider the utility received by a specific customer from the direct channelto be a fraction (with 0 1) of the utility receivedfrom the traditional channel, using for all customers.Similar to their results, we find that the introduction ofa direct channel can sometimes benefit the existingretailers. In contrast to their models, we find thatsupply chain benefits arise from the manufacturer actively participating in the direct channel, as opposedto the mere threat of direct channel sales.Hendershott and Zhang (2004) analyze a modelwith a manufacturer and multiple, heterogeneous retailers. Consumers incur search costs, so in the absenceof a direct channel, they will purchase a product fromthe first retailer that they discover offers a price belowa threshold value. When the manufacturer is able tooffer the product directly to consumers, conditions aredeveloped under which the manufacturer only sellsthrough the traditional channel, only sells through thedirect channel, or sells through both channels. Thehybrid market structure arises only when the direct43channel is more attractive to all customers and thetransaction costs for the web fall within a specifiedrange. This paper does not consider price matching inthe direct channel or customers with heterogeneity intheir channel preference.3.ModelWe consider a supply chain structure where a manufacturer sells to a retailer as well as in a direct channel,similar to the structure of the supply chains in Tsayand Agrawal (2004b), Seifert et al. (2004), and Chianget al. (2003) , but we use a formulation that explicitlymodels the channel preferences of heterogeneous customers. This customer utility model, which we summarize briefly in this paper, is similar to Cattani et al.(2004a), although this research assumes that both theretailer and manufacturer are active players.We model an individual customer with a utilityfunction for a product sold by a firm, where the customer’s utility for the product is decreasing in theprice and the purchase effort (based on the particularchannel) of the product. We use a linear utility function as in Srinivasan (1982) and Vandenbosch andWeinberg (1995). In particular, the customer derivesutility Ui R Ei Pi from the product purchased atprice Pi through channel i, with i 僆 {T, W} for thetraditional and web channels. R represents the reservation price for the product, Ei represents the amountof effort to buy the product in channel i with Ei Unif [0, i] and i is a scaling factor for the channel.We follow the assumptions of Vandenbosch andWeinberg (1995) and assume that R is constant acrossthe population.1 Customer efforts required for eachchannel (ET and EW) are independent and uniformlydistributed. Heterogeneity of the channels is capturedthrough the scaling factors T and W, which can bedifferent for the two channels and are assumed to bestrictly greater than zero.Our model builds on the familiar linear Hotelling(1929) model and captures not only market share, butalso market penetration. In our model, the marketpotential is fixed, but market penetration depends onprice. Thus, demand is elastic and prices can be usedto influence not only the relative market shares, butalso the percentage of the market that purchases aproduct. We model both market share and marketpenetration by assuming that R is sufficiently low sothat not all customers participate. (See Appendix forconditions on R that are sufficient for this assumptionto hold.) This is in contrast to the models in many of1A more complex rendering of this model might assume that Rvaries across the population. Then the net utility would require theevaluation of the sum of the two random variables: effort andreservation prices. If this convolution was uniformly distributed,the resulting model would be equivalent to the current model.

44Cattani et al.: Boiling Frogs: Pricing Strategies for a Manufacturer Adding a Direct Channel that Competes with the Traditional ChannelProduction and Operations Management 15(1), pp. 40 –56, 2006 Production and Operations Management Societythe related works, where full market coverage is assumed. In our model, customers will purchase theproduct only if the net utility from at least one channelis positive. If both channels provide positive utility,the customer will buy from the higher-utility channel.The retailer’s decision is traditional retail price PTwhile the manufacturer chooses both a wholesaleprice B to charge the retailer as well as the web retailprice PW. We first consider the case where the manufacturer commits to matching the traditional channelretail price in the direct channel in order to mitigatechannel conflict (i.e., PT PW P). We analyze therelaxation of this constraint in Section 4.2.The efforts for the two channels for a specific customer (ET and EW) can be compared directly; thesmaller of the two corresponds to the channel that ismore convenient for that customer. The effort consumers require in the traditional channel may be thoughtof as their distance from the retail store. The effortconsumers require in the web channel may be thoughtof as the accessibility and comfort related to buyingvia the web. Consumers with a low web effort may bethought of as experienced web users who value theconvenience of e-tailing, whereas those with a highweb effort may not have convenient web accessand/or may be concerned about the security of conducting financial transactions over the Internet. Weassume that the web effort and traditional effort areindependent.Additional costs to the consumers, such as taxes,shipping or handling costs, could lead to higher “net”retail prices in the two channels. However, our observation is that firms frequently match prices prior tothese costs (see examples in Table 1). While such costscould be incorporated into our model by adjusting therespective channel prices or efforts, it would be at theexpense of tractability and clarity. Also note that thesecosts in practice might roughly offset each other: thefull price the customer will pay in the direct channelmay include shipping and handling, while the fullprice in the traditional channel may include taxes.The relative values of T and W are used to represent different types of populations. Let be the ratio ofthe scaling factors, i.e., W/ T. When 1, thepopulation requires less effort on average to purchasefrom a traditional channel versus the web. One mightexpect that is relatively high for a (relatively new)web channel, but will decrease over time for a particular product as the population in general becomesmore comfortable purchasing products over the Internet. We later demonstrate that the equal-pricing policymay be reasonable for profit maximizing manufacturers when is large but less reasonable as declines.Without loss of generality, we assume that the market size is one. We let CT denote the channel cost perunit that the retailer incurs in selling in the traditionalchannel and CW denote the channel cost per unit thatthe manufacturer incurs in selling in the Internet channel. We assume that manufacturing costs are zero,although all our analytical findings carry through forthe case with non-zero manufacturing cost, with appropriate adjustments to the specific threshold values.Demand curves are determined as follows. Consider first the traditional channel with a manufacturerand a retailer, prior to the manufacturer’s introductionof an Internet channel. The manufacturer is a Stackelberg leader who sets her wholesale price knowinghow the retailer will respond. In particular, it costs theretailer B CT to stock and sell a unit to a customer atprice PT. As described above, the traditional channelfaces demand based on the customer’s distance fromthe retailer. The probability that a randomly selectedcustomer will purchase the product provides a downward sloping demand curve:Pr兵a customer will buy其 Pr兵UT 0其 Pr兵R PT ET 0其 Pr兵ET 共R PT 兲其 共R PT 兲/ T .The traditional retailer chooses a price PT to maximizehis expected profit R: R 共P T 兲 共共R P T 兲/ T 兲共P T B C T 兲.(1)Next consider the addition of an Internet channel.We must now account for effort in the web channel aswell as the effort in the traditional channel. If utility ispositive from at least one channel, the customer willpurchase the product from the channel that providesthe greatest utility. If neither channel provides positive utility, the customer elects not to purchase a product. Expected profit is calculated over the appropriateregions of the joint distribution of ET and EW, therandom variables for scaled traditional and web effort,respectively. For the analysis that follows, we considerthe case that the manufacturer commits to matchingprices in the two channels, i.e., PT PW P. (Weconsider the relaxation of this constraint in section4.2.) Figure 1 illustrates the case where the populationrequires more effort on average to purchase from theweb channel, i.e., 1.While the channel preference is determined by thecustomer’s position relative to the 45 degree indifference line, the purchase decision depends on the interplay of both effort and price, i.e., it depends on the netutility that a customer can receive by purchasing aproduct from the preferred channel. In particular, fora given set of prices, customers with low effort in thetraditional channel (e.g., they live close to the store) orlow effort in the web channel (e.g., they find the webconvenient) are relatively more likely to buy the prod-

Cattani et al.: Boiling Frogs: Pricing Strategies for a Manufacturer Adding a Direct Channel that Competes with the Traditional ChannelProduction and Operations Management 15(1), pp. 40 –56, 2006 Production and Operations Management SocietyFigure 145Customer Channel Preference Based on Scaled Effort.uct, while customers who have high efforts in bothchannels are less likely to purchase the product.Prices above R provide positive utility to 0% of thepopulation while prices below R i provide positiveutility to 100% of the population. We focus on retailprices that satisfy R i P R. Utility is greaterthan zero for all customers whose web effort EW ⱕ (R P) (regions W2, W1, and T1 of Figure 1) and for allcustomers whose traditional effort ET ⱕ (R P) (regions W1, T1, and T2). Customers in region W2 are newcustomers generated through the addition of the webchannel. Customers in regions W1 and T1 receive positive utility from both channels. Customers in regionW1, above the diagonal line ET EW, have a preference for the web channel, while those in region T1have a preference for the traditional channel. For agiven set of parameter values and prices, the fractionof customers in each region is calculated from theeffort distributions as follows:W2 Pr{Customer buys via web but would not buyfrom traditional channel} (1 (R P)/ T)(R P)/ W,W1 Pr{Customer would buy from either channelbut prefers web} (1/2 T W)(R P)2, and T T1 T2 Pr{Customer buys from traditional channel} (R P)/ T W1.3.1. Base Case: A Single ChannelIn this section, we determine optimal wholesale andretail prices for the traditional channel prior to theintroduction of the Internet channel for comparison asa base case, and we use the subscript b to identifyprices and resulting profits for this case. It is easy toshow that, given any wholesale price Bb, the retailer’sprofit maximizing retail price is Pb*(Bb ) (R CT Bb)/2.Given that she knows how the retailer will respond,the manufacturer chooses the profit-maximizingwholesale price Bb* (R – CT)/2, leading to the basecase retail price Pb* (3R CT)/4. Resulting expected* (R CT)2/(8 T) forprofits in the base case are bM* (R CT)2/(16 T) for thethe manufacturer and bRretailer.As a point of reference, if the channels were centrally owned, expected supply chain profits would be(R CT)2/(4 T). Thus the effect of double-marginalization is a reduction in expected supply chain profitsof (R CT)2/(16 T).3.2.Manufacturer Adds a Direct Channel WhileEnsuring Prices in the Direct Channel MatchPrices in the Traditional ChannelIn this section, we analyze the effect of the manufacturer’s introduction of an Internet-based channel onthe supply chain. A naı̈ve strategy for both the manufacturer and the retailer would be to continue to usewholesale and retail prices as in the base case. Whilewe suspect that the naı̈ve strategy is used at leastoccasionally in practice, it is clearly suboptimal. Thus,here we assume that the manufacturer, as a Stackelberg leader, updates her wholesale price knowinghow the retailer responds given the supply chainstructure with its two channels. We consider threestrategies for the manufacturer, each of which assumes that the manufactur

Sony 32 Flat Tube HDTV 1,299.99 1,299.991 Sony 57 HD Projection TV 2,299.99 2,299.991 Adidas Climacool 2 M Shoes 100.00 99.992 Nike Air Kantara Shoes 120.00 119.992 Nike Triax Stamina Watch 129.00 129.002 LEGO Snowboard Superpipe 49.99 49.233 1Best Buy. 2Dick’s Sporting Goods. 3Wal-Mart.