Changing Channels

Many simple and effective methods exist for measuring the profit impact of three of the four P's of marketing: price, promotions, and product. However, controlled experiments in test markets are typically unavailable for the fourth P: place (distribution). New research helps predict the results of proposed changes in distribution channels.

In the study "Assessing the Economic Value of Distribution Channels," University of Chicago Graduate School of Business professor Pradeep K. Chintagunta, Junhong Chu of the business school at the National University of Singapore, and Naufel J. Vilcassim of London Business School examined manufacturers' distribution channel and product line options in the PC industry.

The movement of products from manufacturer to consumer is referred to as a "distribution channel," with agents en route to the consumer serving as intermediaries. Firms in industries selling multiple products and using multiple distribution channels (such as the PC industry) have to decide which distribution channels to use and which product to sell through which distribution channel. Distribution strategies are intertwined with a firm's product line, market segmentation, and positioning and targeting strategies. Changes to distribution strategies affect a firm's customer base and profits, as well as its relationships with intermediaries and other firms in the market.

Chintagunta, Chu, and Vilcassim developed a statistical model that can help managers determine which distribution channel works best and evaluate the financial implications of changing the firm's distribution strategy and channelproduct line combinations.

"It's difficult to predict the result of changes in distribution channels, because existing relationships with intermediaries prevent firms from conducting actual experiments to test new strategies," explains Chintagunta.

Gauging the outcomes of new distribution strategies is especially important because contracts between manufacturers and retailers are long-term relationships and cannot arbitrarily be broken.

"Modifying distribution channels is an expensive, timeconsuming task that leaves many people upset, notably intermediaries who might be 'dis-intermediated,'" says Chintagunta. "In those contexts, a firm needs to have better guidance regarding the consequences of their actions."

A Changing Industry

"There were two key triggers that changed distribution strategies in the PC industry," explains Chintagunta. "As Dell and direct sellers became more powerful and the Internet appeared in the mid-1990s, larger companies like HP and IBM began migrating to the direct channel. This shift enabled large manufacturers to sell and distribute their products directly to customers. The Internet provided a good catalyst for thinking about how to restructure distribution channels."

Traditional retail channel margins account for 20 to 30 percent of a product's price. From a manufacturer's perspective, if they can sell as well as intermediaries, selling directly to consumers is much more profitable; the manufacturers will not have to give a share of profits to channel partners. Eliminating the retail markup via direct selling also benefits consumers in terms of lower costs.

To study the result of changing distribution strategies, Chintagunta, Chu, and Vilcassim developed a "structural model" that simulates marketing policy changes. The model specifies the economic agents being studied (consumers, retailers, and manufacturers) and the nature of their interactions. The authors then used the model to predict the behavior of each set of agents in response to a manufacturer changing an element of its distribution strategy.

"We used data from the PC industry to determine the parameters of the model, but eventually we needed the model to predict the effects of policy changes," says Chintagunta. "Our statistical model characterizes the interactions between sales and prices of all products sold in all channels, allowing us to assess the economic value of each channel to each firm and its consumers."

The structural model allowed the authors to conduct a "What if?" type of analysis that could be used as an input for marketing managers.

The authors used data from the U.S. PC market from 1995 to 1998, including retail prices and unit sales at the manufacturer's brand-model level across six different distribution channels. At the top of the channel are such manufacturers as Dell, Compaq, and Gateway. Each manufacturer has several brands, each of which has a series of PC models. Together, these models form a product line. For example, Dell has the Dimension brand, and multiple products within this brand compose the Dimension product line.

The six distribution channels evaluated in the study are:

1. Direct outbound: sales by a manufacturer's sales team
2. Direct inbound: telemarketing or catalog sales
3. Dealer: corporate account resellers and computer specialty dealers, usually focused on sales to large volume buyers
4. Retail: storefront businesses that sell to a large number of unrelated consumers
5. Internet: direct sales via manufacturer Web sites
6. Other

Indirect channels (dealer and retail) accounted for the bulk of PC sales, with 35 percent of PCs sold via dealers and 31 percent sold in the retail channel. Internet sales accounted for less than 1 percent of total PC sales in 1998.

All manufacturers used multiple channels to market their products, but different manufacturers focused on different channels. Each product line had one primary channel. These product line/channel arrangements help firms match consumer preference for the product line with preference for the channel. This, in turn, helps reduce intra-brand competition across channels and better coordinates the manufacturer's distribution strategy. For example, Compaq sold 90 percent of its Presario PCs via the retail channel, 85 percent of HiNote PCs via dealers, and 99 percent of ProSignia PCs via catalogs and its Web site.

"We took a statistical modeling approach that had primarily been used in economics and extended it to a mainstream marketing problem of matching products with distribution channels," says Chintagunta.

Experimenting with Distribution Strategies

To assess the economic value of a distribution channel to a PC manufacturer, the authors dropped the distribution channel from a manufacturer's distribution system and recalculated average prices and market shares under the new distribution system. The authors also recalculated changes in the manufacturer's profits.

Distribution channels with high historical sales and profits were not necessarily the most economically valuable to firms. In addition, there were considerable differences in channel substitution patterns. For example, there was little substitution between direct and indirect distribution channels.

The removal of any distribution channel benefits rival PC manufacturers because it reduces competition. Removal of direct channels will benefit intermediaries, but removal of indirect channels will hurt them. In all cases, consumers are worse off when a distribution channel is dropped.

Manufacturers usually incur losses when their product lines are dropped entirely from some channels. Rival manufacturers always benefit from the removal of a manufacturer's product line because it lowers competition. Intermediaries benefit if a product line is dropped from a direct channel because some of the previous purchasers of the product line will switch to indirect channels, but intermediaries will be hurt if a product line is dropped from an indirect channel due to lost profits. Removal of a product line negatively impacts consumers because it decreases the amount of variety in the market.

To study the effect of adding channels to a firm's distribution systems, the authors simulated various strategies for Dell, Compaq, and Gateway.

Dell entered the retail channel in 1990 but exited in 1994 due to heavy losses. Through policy simulations, the authors found that if Dell switched any of its current product lines from direct channels to retail channels, it would be worse off by several million dollars. Adding product lines from direct channels to retail channels slightly increases Dell's profits, but if these profits cannot recover the fixed costs associated with entering the retail channel, Dell will be better off not entering offline retailing.

One of the major strategy shifts for Compaq in the late 1990s was a shift to direct selling. The authors find that switching product lines from retail channels to the Internet or extending product lines from dealer to direct inbound would increase Compaq's profits substantially. But while Compaq's decision to sell direct was economically justified, the authors suggest that the firm should have taken measures to coordinate with its channel partners due to substantial losses for the intermediaries.

Gateway's major channel strategy shift was to close all its retail outlets and use third-party retailers. According to the authors' estimates, Gateway would have higher profits by switching retail sales to the Internet.

Consumers are better off when a product line is added to another channel because they will have more choices. They also benefit when product lines are switched from the retail channel to the Internet. However, consumers are worse off when product lines are moved from direct channels to retail channels.

The authors next studied the HP-Compaq merger, because one motivation for the merger was to change the firms' distribution strategies. The $25 billion merger in 2002 was considered the largest deal in PC industry, creating a powerhouse firm that could compete with IBM and potentially develop a business structure similar to Dell.

The authors simulated the potential impact of the merger on all parties involved (the new company, rival firms, and consumers) under various post-merger channel strategies.

"Our findings are consistent with what HP and Compaq discussed; the companies would have been much better off by selling directly to consumers," says Chintagunta. "However, distributors' profits would have decreased dramatically. The changes in the distribution strategy resulting from the merger might have been better managed if the firms could have predicted the economic impact on their channel partners. This is where a structural model such as ours comes in."

New Research Tools

Chintagunta, Chu, and Vilcassim's study is significant for firms considering restructuring their distribution channels, as well as rival firms interested in understanding the impact of a channel mix change by a competing manufacturer. The model can be applied to other industries with channel structures similar to the PC industry.

A potential direction for future research is examining the long-term impact of Internet sales. The authors note that their conclusions about the Internet channel are necessarily limited because their data set ends in 1998, whereas, the Internet did not become a major sales channel for the PC industry until 1997.

"Firms should develop some familiarity with structural modeling," says Chintagunta. "Market research groups are capable of implementing these types of models to develop reasonable predictions before making major changes in their distribution strategies."