number 10 number 23

number 11 number 23

number 11 number 23

number 11 number 23

number 11 number 23

number 11 number 23

Car companies are understandably focused on increasing customer loyalty, since it can mean a big boost in profits.

Traditionally, automakers have relied on data from survey firms like J.D. Power to help them understand why customers stick with a vehicle brand. But according to new research from Seethu Seetharaman, a professor of marketing at Rice University’s Jesse H. Jones Graduate School of Management, the picture that type of data paints can be incomplete, since it doesn’t consider how quickly consumers replace their cars. Seetharaman proposes a new approach to assessing brand loyalty, one that includes a measure of future market share.
 
Companies in just about every industry go to considerable lengths to boost customer loyalty. And with good reason: numerous studies have shown that even small improvements in customer loyalty can have a significant positive impact on profits. Few industries, though, focus as much attention on strengthening brand loyalty as the automotive business, where profit margins for individual cars can be large.
 
But the auto business faces a unique challenge in trying to encourage customers to remain loyal: the relatively long period of time between car purchases. “As soon as a consumer buys a car, it’s safe to assume that the probability of them repeat-purchasing a car in the next month is as close to zero as you can get,” said Seethu Seetharaman, a professor of marketing at Rice University’s Jesse H. Jones Graduate School of Management. That fact has an important implication for automobile marketers. “Knowing how far into that purchase cycle the customer is tells the marketer how much solicitation the marketer should send to this customer.”
 
Obviously, car companies are aware of this dynamic and they attempt to get as accurate a picture of consumer loyalty as possible through numerous tools, including the well-known J.D. Power Automotive Survey, which compiles huge amounts of data, including how many people bought specific car makes and models each year. But in his groundbreaking new research study, “Speed of Replacement: Modeling Brand Loyalty Using Last-Move Data,” Seetharaman points out some deficiencies in the J.D. Power data while offering a new way to gauge brand loyalty.
 
“To the best of my knowledge, J.D. Power has never reported any metrics speaking to whether a particular model and make gets replaced any faster than another model and make,” said Seetharaman, who conducted the research with Hai Che, an assistant professor of marketing at the University of California’s Haas School of Business. “Even if fewer people bought a certain model this year, if one could show that systematically all those buyers repeat-purchase sooner, then perhaps that bodes well for the manufacturer.”
 
Using his expertise in hazard functions — probability functions that help researchers understand how events occur in time — and by taking into account what’s known as “last move data,” which helps account for how quickly a consumer replaces a car, Seetharaman was able to develop a model for estimating brand loyalty. His model takes into account two factors: how many people make repeat purchases and how quickly they do it. With this approach, Seetharaman and Che found that, at least during their period of study, Ford had the highest brand loyalty, followed by General Motors and Chrysler.
 
The problem, Seetharaman said, is that their approach yielded the exact same results as J.D. Power’s methodology, which simply tallies the number of people repeat-purchasing GM, Ford, and Chrysler cars. “We asked ourselves, OK, how do we convince managers to adopt this approach? Is there something else we could illustrate that could convince the managers that there is something here?” he said. In fact, there is something: Seetharaman’s approach can help predict market shares over the next 12 to 24 months, crucial information because auto companies base their production schedules around those kinds of projections.
 
The key to being able to come up with those projections, of course, is understanding replacement times, something Seetharaman’s approach enables. “That information should be exploited to now predict market shares over the next two or three years, and those market shares will vary over time,” he said. Seetharaman cited the example of a make and model of a car that is replaced every seven years or so. “There’re a lot of people who bought this make and model seven years back who are all now in the market for a repeat purchase. You are going to see a spike for that make and model this year,” he said.
 
A prediction model that doesn’t take into account replacement times, including how they vary from person to person, is not going to be able to make that kind of projection. “It’s going to make a market share prediction that is kind of an average prediction based on how many people currently own a brand,” he said. “There will be no basis to predict whether the market share will be higher in 2009 than in 2008. You have to take into account these heterogeneous replacement patterns.”
 
For more information, contact Seethu Seetharaman at seethu@rice.edu or Laura Hubbard of the Jesse H. Jones Graduate School of Management at lhubbard@rice.edu.