Knowledge that Transforms

The question as to the optimality of advertising pulsing has attracted many researchers over the last half-century. In this paper we specify a market share model in which there are two advertising-setting firms as well as a no-purchase option. The framework is that of a first-order Markov process with three states. The objective of both firms is to maximize profits. We are able to demonstrate, for a diminishing returns advertising function, that the optimal advertising strategy is pulsing. The frequency of the advertising pulse is shown to depend on the magnitude of the market share retention rate (state dependence); the higher it is, the less frequent the advertising. We further find that the optimal advertising budgets do not remain the same when the frequency of pulsing changes. Finally, we show that it is optimal for both firms to advertise in phase.

This paper reports on a large-scale implementation of marketing science models to solve the bidding problem in search engine advertising. In cooperation with the online marketing agency SoQuero, we developed a fully automated bidding decision support system, PROSAD (PRofit Optimizing Search engine ADvertising; see http://www.prosad.de), and implemented it through the agency's bid management software. The PROSAD system maximizes an advertiser's profit per keyword without the need for human intervention. A closed-form solution for the optimized bid and a newly developed “costs-per-profit” heuristic enable advertisers to submit good bids even when there is significant noise in the data. A field experiment demonstrates that PROSAD can increase the return on investment by 21 percentage points and improve the yearly profit potential for SoQuero and its clients by €2.7 million.

Patients increasingly request their physicians to prescribe specific brands of pharmaceutical drugs. A popular belief is that requests are triggered by direct-to-consumer advertising (DTCA). We examine the relationship between DTCA, patient requests, and prescriptions for statins. We find that although the effect of requests on prescriptions is significantly positive, the mean effect of DTCA on patient requests is negative, yet very small. More interestingly, both effects show substantial heterogeneity across physicians, which we uncover using a hierarchical Bayes estimation procedure. We find that specialists receive more requests than primary care physicians but translate them less into prescriptions. In addition, we find that the sociodemographic profile of the area a physician practices in moderates the effects of DTCA on requests and of requests on prescriptions. For instance, physicians from areas with a higher proportion of minorities (i.e., blacks and Hispanics) receive more requests that are less triggered by DTCA and are accomodated less frequently than physicians from areas with a lower proportion of minorities. Our results challenge managers to revisit the role of DTCA in stimulating patient requests. At the same time, they may trigger public policy concerns regarding physicians' accommodation of patient requests and the inequalities they may induce.

We show how networks modify the diffusion curve by affecting its symmetry. We demonstrate that a network's degree distribution has a significant impact on the contagion properties of the subsequent adoption process, and we propose a method for uncovering the degree distribution of the adopter network underlying the dissemination process, based exclusively on limited early-stage penetration data. In this paper we propose and empirically validate a unified network-based growth model that links network structure and penetration patterns. Specifically, using external sources of information, we confirm that each network degree distribution identified by the model matches the actual social network that is underlying the dissemination process. We also show empirically that the same method can be used to forecast adoption using an estimation of the degree distribution and the diffusion parameters at an early stage (15%) of the penetration process. We confirm that these forecasts are significantly superior to those of three benchmark models of diffusion. Our empirical analysis indicates that under heavily right-skewed degree distribution conditions (such as scale-free networks), the majority of adopters (in some cases, up to 75%) join the process after the sales peak. This strong asymmetry is a result of the unique interaction between the dissemination process and the degree distribution of its underlying network.

It is well known that individuals often fail to exert proper self-control. In organizational settings, this can lead to reduced productivity and profits. We use the literature on present-biased preferences to model employees' self-control problems and examine how firms can design compensation plans to reduce the negative consequences of their employees' self-control problems. Our results suggest that firms can mitigate self-control problems by delaying payment to the employees. This can be achieved by using multiperiod quotas (such as annual quotas) to compensate employees for their cumulative performance. Although such plans are prevalent in the market, there is little theoretical research that shows when multiperiod quota plans can be optimal. The paper provides one potential explanation for the widespread use of such quota plans. Interestingly, we find that such plans may be optimal despite the fact that they encourage more procrastination. We also find that such plans lead to higher effort by the employees and can sometimes improve the welfare of not only the firm but also the employees.

There exists a dichotomy in the communication strategies of fashion firms—some firms purposefully cloak information on the tastefulness of their products, whereas others openly flaunt their tasteful or “it” products. This divide in communication strategies cannot be explained by existing wealth signaling models of fashion. In this paper, we offer a model of fashion that explains the above dichotomy. We model fashion as a social device that plays the dual role of allowing people to both fit in with their peers and differentiate themselves by signaling their good taste or access to information. In this context, we show that a fashion firm faces an interesting dilemma—if it restricts information, then only sophisticated consumers buy its products and use them to signal their taste. Cloaking thus preserves the signaling value of its products but reduces the number of social interactions enabled by them. In contrast, flaunting undermines the signaling value of its products but increases the interactions enabled by them. Given these trade-offs, we derive the conditions under which cloaking occurs. We also show that, in equilibrium, the most tasteful product endogenously emerges as the fashion hit or “it” product.

Firms that sell via a direct channel and via indirect channels have to decide whether to allow third-party sellers to use trademarked brand names of products in their advertising. This question has been particularly controversial for advertising on search engines. In June 2009, Google started allowing any third-party reseller of a product to use a trademark such as “DoubleTree” in the text of its ad, even if the reseller did not have the trademark holder's permission. We study the effects of this change empirically within the hotel industry. We find some evidence that allowing third-party sellers to use a trademark in their online search advertising weakly reduced the likelihood of a consumer clicking on a trademark holder's paid search ads. However, the decrease in paid clicks was outweighed by a large increase in consumers clicking on the unpaid links to the hotelier's website within the main search results. Our evidence shows that when a third-party seller focuses on a trademarked brand in its ads, the ads become less distinct, and customers are more likely to ignore the advertised offers and buy from the direct channel.

This paper asks whether brand extension can serve as a signal of product quality given that it costs less than a new brand. (Existing literature has assumed either that brand extension is cost-neutral or that it costs more.) I show that it can as a perfect Bayesian equilibrium, but the argument is unconvincing. For one thing, the separating equilibrium is not unique; a pooling equilibrium also exists in which brand extension signals nothing. For another, the separating equilibrium relies on off-equilibrium beliefs that are poorly motivated in the model. I propose a refinement of the perfect Bayesian equilibrium that resolves both issues. Empirical off-equilibrium beliefs require that consumers' off-equilibrium beliefs be justifiable on the basis of their prior beliefs and product performance observations. With empirical off-equilibrium beliefs, two necessary conditions for brand extension to signal product quality are identified: (i) consumers must perceive old and new products of the firm to be positively correlated in quality, and (ii) at least some consumers must identify with brands and not the firm behind the brands. Even with these conditions in place, the signaling argument is fragile: firm observability of past performance diminishes brand extension's signaling capability; an arbitrarily small probability of failure for good products eliminates it. My results suggest that, going forward, the case for brand extension must rest on foundations other than signaling product quality.

The goal of this paper is to study the behavior of consumers, dealers, and manufacturers in the car sector and present an approach that can be used by managers and policy makers to investigate the impact of significant demand shocks on profits, prices, and dealer networks. More specifically, we investigate consumer demand, substitution patterns, and price decisions across different cars and dealer locations to identify dealerships with low margins or high fixed costs and measure the value of closing down dealerships for manufacturers. We apply our model empirically to the San Diego area using a transactional data set with information about the locations of dealers and consumers, as well as manufacturer and retail prices. We find strong consumer disutility for travel and find that dealers have local demand areas that are shared with a small set of competitors. We show that a reduction of market demand by 30% over two years, similar to the economic crisis of 2008–2009, results in an annual drop in prices of approximately 11%. We discuss this price drop in the context of the 2009 federal policy measure known as the Car Allowance Rebate System program. We compare predictions and actual dealership closings in the General Motors and Chrysler dealer networks as an application of our approach.

Volunteer users employ collaborative Internet technologies to develop open source products, a form of user-generated content, where time to product release is a crucial measure of project success. The open source community features two separate but related subcommunities: developer users who contribute time and effort to develop products and end users who act as collaborative testers and provide feedback. We develop hypotheses concerning how the location of the project's founders in the social network of developer users, the interplay of developer users and end users, and project and product characteristics affect time to product release. We use data on 817 development projects from SourceForge, a large open source community forum, to calibrate a split hazard model to test the hypotheses. That model supports the two-community conceptualization and most of the related hypotheses. The results have theoretical and managerial implications; for example, a pivotal position of founders in the developer user community can reduce time to product release by up to 31 and projects in which users are more engaged can experience an 11 time to product release compared with those projects in which they are not.