A method for measuring the 'cool factor' of commercial products

Computational model measures consumer choices in terms not only of price and usefulness, but also network effects

Whether buying shampoo, cars, or clothing, consumers ponder more than price and usefulness when deciding how to spend their money.

So-called "network effects"—the term for the value derived from the product's popularity within a large community of users—also come into play.

A Johns Hopkins University business professor has developed a computational model that measures consumer choices in terms not only of price and usefulness, but also of network effects. The findings of the new study, which will be published in the journal Management Science, could be valuable to manufacturers and retailers seeking to boost sales and market shares.

"Suppose you're planning to buy a computer. You'll consider a price range that you're comfortable with, and you'll look at all the attributes of the computer—the screen, keyboard, memory, CPU, and so on," says Ruxian Wang, an assistant professor at the Johns Hopkins Carey Business School and the paper's lead author. "But you could also consider the network effects. Specifically, is this a popular product that makes you feel happy to be among the people buying and using it? The model described in this paper computes data in a way that shows the extent of the network effects, and it can accurately predict the future sales of a product based on how strong or weak its network effect is, in addition to its features or price promotion."

An example of a positive network effect—the cool factor associated with, say, a fashion trend or a ticket to Hamilton, Broadway's hottest show. But, Wang warns, this phenomenon cuts both ways: When so many people are buying the current trendy footwear or dress design that it begins to appear passé, a negative network effect may arise. The same would apply to a commuter route that appeals to so many drivers that it leads to snarled traffic.

Knowing the network effects of products can help companies determine how many of these items it should make, says Wang. Instead of offering an array of similar products—say, a line of computers or automobiles that are more alike than different—a company could use Wang's algorithm to determine which of the products had strongly positive network effects and then concentrate on selling those. Wang cites the example of Apple and its relatively limited but highly successful line of technology products.

"In the past, selling a variety of items was usually the way firms approached their work," Wang says. "But that can be very expensive for them. There's really no reason to offer many different versions of a particular product if the company finds that one or two of them have strong network effect. They could even lower the price of that product a little and experience an increase in sales and a larger share of the market."

Streamlined product lines also can reduce the confusion that potential customers sometimes feel when faced with a large number of similar items, according to Wang.

The study, titled "Consumer Choice Models with Endogenous Network Effects," was co-authored by Zizhuo Wang, assistant professor at the University of Minnesota,. The computational model developed by the two authors was based on data from downloads, player ratings, and other user information from the Google platform for video games. Wang points out that this study was concerned with a single firm and how it would market multiple products. He and his colleague plan in future studies to consider how network effects influence the strategies of competing manufacturers and retailers.