How to Sell Products on the Web

Research by Wendy Moe

Researchers have long believed that consumers make purchase decisions in two stages. In the first stage (stage 1), consumers select a smaller group of products from the wide choice that is available to them. In the second stage (stage 2), consumers select and purchase a product from this smaller set. Because it is very difficult to observe the first stage, researchers have several theories about consumer behavior in that stage but little empirical evidence. Dr. Wendy W. Moe, assistant professor of marketing at the Smith School, in her paper “Empirical 2-Stage Choice Model with Varying Decision Rules Applied to Internet Clickstream Data,” examines both these stages using data from online purchases.

Moe’s research shows that consumers are looking at different product attributes in the two stages. Consumers also tend to do a simpler mental analysis in stage 1 compared to stage 2. Moe’s study also reveals that there may be some preliminary evidence that consumers consider product attributes such as price and size only in one of the two stages; whereas other product attributes are considered in both stages. From a managerial perspective, Moe’s results have strong implications for product marketing strategies.

Moe collected seven months of data from a retailer of nutritional products. She then focused her analysis on the two most popular products categories on the Web site: weight loss aids and meal replacements. She looked at data from the purchasing activities of 142 buyers in the weight loss aid category and 143 buyers in the meal replacement category. Buyers in the weight loss aid category accounted for 420 stage 1 and 278 stage 2 choices. Buyers in the meal replacement category accounted for 438 stage 1 and 295 stage 2 choices. The data she collected captured the specific page requests made by shoppers. For weight loss aids she identified six attributes: price, size, brand, ephedrine, caffeine, and St. John’s Wort. For meal replacements, she identified five attributes: price, size, brand, flavor, and protein.

Moe found that in stage 1, where consumers are screening the wide choices available to them to select a smaller set, they tend to use simpler mental decision rules than in stage 2, where they are making purchase decisions on the products in the selected smaller set. Price and size tended to be important in only one of the two stages, whereas ingredient attributes play a role in both stages. Flavor tended to be more important in the second stage than the first stage.

The research findings have strong implications for promotional design and targeting. From a managerial standpoint, it is important to identify which product attributes are being used for screening decisions (stage 1), for purchasing decisions (stage 2) and for both. This information combined with the knowledge of how a product is performing in the two stages can enable managers to devise better targeting strategies. Also, if consumers show a high preference for variety in stage 1, it possibly indicates greater consumer uncertainty and therefore perhaps a need for more education in the form of product information.

Although these results may be applicable to other products, Moe is quick to point out that her analysis was done on weight loss aids and meal replacements. Thus, it is not entirely certain that analysis of other products will reveal similar results.

Moe recommends that marketing managers identify the product characteristics that are important in each stage, and then promote their products based on how the product is performing in each stage. Managers should look at the strengths and weaknesses of the product in each stage and then select an appropriate intervention.

The availability of Internet commerce data has opened up the possibility of a flood of related research. Moe plans to continue to explore the data-rich world of Internet shopping. “I am interested in tracking a customer right from the point that they visit the Web site to the point that they make a purchase or no-purchase decision,” says Moe. “Eventually, I hope my research generates a model which allows me to make probabilistic predictions as to whether a customer will buy something or not.”

“Empirical 2-Stage Choice Model with Varying Decision Rules Applied to Internet Clickstream Data” was published in the Journal of Marketing Research. For more information about this research, please contact wmoe@rhsmith.umd.edu.