Introduction to Theory and Practice in Marketing Conference Special Section of Marketing Science

Introduction

Over the last two decades, we have witnessed a dramatic shift in how consumers search for information, consume media, connect with each other, and buy products. This shift has had a profound impact on the marketing practices and business models of firms. It is an exciting time for marketing scholars to explore and investigate a series of new and impactful phenomena. Many industry experts believe that the world of marketing is turning upside down: traditional advertising industry is dramatically changing as a result of disruption by ad exchanges; brands are becoming publishers and entertainers, blurring the line between the content and message; and “big data” in real time are becoming important sources for customer insight.

Marketing academics address such new marketing topics, as well as the more traditional ones, using a wide variety of methodological tools. Most academic papers are motivated by practical marketing problems, and the review process at Marketing Science and similar journals invariably evaluates papers based on their relevance. However, the relevance and impact of academic research is limited if it reaches only other academics or research specialists within companies, not line managers or senior decision makers.

In part as a result of the growing complexity and specialization of academic research, the issue of relevance and impact on practice has become an ongoing debate at business schools worldwide. Many observers complain that academia is far removed from addressing substantive problems of industry. This perception threatens to make our field irrelevant and undermine the legitimacy of research at business schools.

This special section is an outgrowth of an effort by many people to nudge our field toward bridging the gap between theory and practice. We hope this effort complements other initiatives, such as the Science-to-Practice Initiative (Desai et al. 2012), the Gary L. Lilien ISMS-MSI Practice Prize awards papers in Marketing Science,¹ efforts by the board of the INFORMS Society of Marketing Science (ISMS), and the work of the Marketing Science Institute (MSI).

Theory + Practice in Marketing (TPM)

The Origin

In early 2011, Bernd Schmitt and Don Lehmann at the Columbia Business School and Sunil Gupta at the Harvard Business School began informal conversations about the lack of relevant research and the overemphasis on sophisticated methods in our field. They coined the term “TPM” (for “Theory + Practice in Marketing”) and organized a one-day symposium at Columbia University. Approximately 40 influential marketing scholars were invited; in total, 70 leading experts in our field attended the symposium, including current and former editors of major journals and three deans of the leading business schools.²

Almost everyone at this symposium shared the frustration about the limited relevance of our research. We wondered why we hesitate to bring much of our research into our MBA classrooms and why we don't feel comfortable talking about our research to executives.

The journal editors explained that although they encouraged and wished to see more relevant research published, their hands were tied—reviewers focused almost exclusively on the rigor or sophistication of methods. In many cases, reviewers rejected papers that emphasized substantive issues with simple methods. We received a broad endorsement to continue TPM to represent a voice about the latent desire among marketing scholars for relevance.

The Follow-up Conferences

For the second TPM conference, which was held at the Harvard Business School in 2012, we decided to invite interested scholars to share their relevant research. Preyas Desai, the editor of Marketing Science, approached us to support this effort by offering a special section of Marketing Science for papers accepted from this TPM conference. Marketing Science's goals of providing substantive insights and making a strong impact on practice (e.g., Ratchford 2001, Desai 2011, Chintagunta et al. 2013) and its openness to a wide of variety of methods made the alliance with the TPM conference an easy decision. Dominique Hanssens of the University of California, Los Angeles and John Hauser of the Massachusetts Institute of Technology joined the team for the conference and special section. Decisions were difficult: we received more than 80 papers for the conference but could accommodate only about two dozen for presentations. All invited papers became eligible for the special section in Marketing Science—they would go through a rigorous review process with a special eye for their practical relevance. The five papers included in this special section are the outcome of this effort.

Nader Tavassoli organized the third TPM conference in April 2013 at the London Business School. This time, we were glad to receive the support of Bob Meyer, the editor of the Journal of Marketing Research (JMR), to publish a special issue based on the papers presented at this TPM conference. Once again we received an enthusiastic response. Researchers submitted almost 100 papers, of which about three dozen were selected for presentation and became eligible for the special issue of JMR. Greg Carpenter will host the fourth TPM conference in 2014 at the Kellogg School of Business, and V. Kumar from Georgia State University will host the fifth TPM conference in 2015. We hope this joint effort will nudge our field toward more relevant and impactful research.

Special Section of Marketing Science

In the spirit of TPM, the five papers published here address issues of theory and practice. The focus of these papers is on relevant phenomena and the right method, be it simple or sophisticated, to address such phenomena. All papers went through a rigorous review process to ensure that they use sound approaches and draw conclusions based on strong evidence.

In the first paper, “Manufacturer and Retailer Strategies to Impact Store Brand Share: Global Integration, Local Adaptation, and Worldwide Learning,” Jan-Benedict E. M. Steenkamp and Inge Geyskens examine why store brand (SB) shares vary dramatically by country (e.g., average SB share in the United Kingdom is 46%, whereas it is only 4% in Japan and 11% in Italy) and category (e.g., SB share for fruit juice in the United Kingdom is 82% versus 22% in the United States, but the SB share for tea in these two countries is 18% and 14%, respectively). The authors examine the implications for global or local marketing strategies of national brands and retailers. Using three to five years of scanner data and survey data from more than 20,000 consumers and scores of categories from 23 countries, they find that some factors (e.g., innovation) can be part of manufacturers' global integrated strategy, whereas others (e.g., price promotion) should be locally adapted. Some factors (e.g., advertising) should be part of firms' worldwide learning strategy because their effects differ dramatically but predictably across countries.

In the second paper, “Morphing Banner Advertising,” Glen L. Urban, Guilherme (Gui) Liberali, Erin MacDonald, Robert Bordley, and John R. Hauser suggest an approach to morph banner ads dynamically to match the latent cognitive styles (e.g., impulsive-analytic, deliberative-holistic) of consumers. Using responses from more than 100,000 consumers and over 450,000 ads, they test their approach on the CNET website and show that real-time morphing of ads can improve clickthrough rates (CTRs) of banner ads by 83%–97% compared with existing approaches. In a second test with an automotive company, they show that banner ad morphing increased CTRs by 66%– 245%, and even after four weeks of ad exposure, brand consideration and purchase intentions for cars increased by 30% and 8%, respectively. This powerful approach works on a large-scale website in real time and dramatically improves on relevant metrics (CTR, brand consideration, purchase intention).

In the third paper, “A Theory for Market Growth or Decline,” Steven M. Shugan and Debanjan Mitra propose a provocative theory to explain and predict the growth and decline of markets. Using Darwin's natural selection mechanism as the basis, they suggest that environmental mutations cause markets to change. A key testable hypothesis from this theory is that greater displacement precedes growth and stability precedes decline. Shugan and Mitra create a metric of displacement and test their hypothesis using two data sets. They show that their theory better predicts market growth or decline than other benchmark models that include covariates such as trend, past sales, and economic indicators. This insight can prove to be influential for firms considering either entry or exit in certain categories, depending on their outlook for growth.

In the fourth paper, “Outsourcing Retail Pricing to a Category Captain: The Role of Information Firewalls,” Vincent R. Nijs, Kanishka Misra, and Karsten Hansen explore the public policy implications when a manufacturer becomes a category captain (CC) for a product category to provide advice to retailers. Although this relationship may help retailers learn from the category expertise of manufacturers, the Federal Trade Commission (FTC) is worried about implicit collusion between manufacturers and retailers that may have an adverse impact on social welfare. As a result, the FTC recommends strong firewalls—vertical firewalls between the teams that set manufacturer wholesale prices and the teams that suggest retail prices, and horizontal firewalls between teams that suggest prices for competing retailers. Using data from multiple manufacturers and retailers, Nijs et al. estimate a structural model and show that CC arrangements enhance social welfare when vertical firewalls are removed but horizontal firewalls are enforced.

In the fifth paper, “Predicting Individual Behavior with Social Networks,” Sharad Goel and Daniel G. Goldstein of Microsoft Research ask whether social network data improve the prediction of individual-level behavior beyond currently used demographic or behavioral models. Using data from over 100 million users from Yahoo!'s communication network across multiple domains, they find social data improve the prediction of individual behavior and targeting capabilities of companies in almost all cases. The exception is when detailed transactional data are available. This is a nice example of using a large real-life data set across multiple settings to arrive at generalized findings and boundary conditions.

A View from Practice

At the second TPM conference at Harvard in 2012, where the five papers just described were presented, we were delighted to have the following senior practitioners join us to share their perspectives:³

• A. G. Lafley, former Chairman of the Board and CEO, Procter & Gamble (P&G);

• Ajay Banga, CEO of MasterCard;

• Eugenio Minvielle, President and CEO of Unilever, North America;

• Jeffrey Severts, Chief Marketing Officer, Best Buy Europe;

• Chris LaSala, Director of Mobile Partnerships, Google;

• Wes Nichols, Cofounder and CEO, MarketShare;

• Steven Cohen, Partner and Cofounder, In4mation Insights; and

• Vipin Mayar, Senior Vice President, Customer Knowledge and Insights, Fidelity Investments.

While industry leaders highlighted issues of relevance to their specific industry or business, several common themes emerged from their presentations and discussions. These themes suggest potential research projects of high relevance:

1. Competition: How does one compete in a world with almost perfect information and low switching cost for consumers? Severts wondered how retailers such as Best Buy could compete with Amazon and the “showrooming” effect. Minvielle explained that mainstream brands are finding it hard to compete with the premium and discount brands and wanted to know if these brands should think of competition differently.

2. Brands: Lafley suggested that academics should study how firms can grow small brands and resurrect old brands such as Old Spice. Nichols wondered if brand building still matters when consumers can compare prices in real time. LaSala wanted to know how brands should be built across various touch points and what the role of mobile advertising in brand building should be.

3. Consumers: Mayar asked how one could engage consumers online when their attention span is short. Lafley also wanted to know how firms could have a dialogue with consumers in real time. He suggested that research should be iterative and in real time instead of using traditional methods of collecting large amounts of data, taking a sample, and analyzing it after several weeks or months. Minvielle argued that one should understand the rituals of buying and consuming products, whether it is for a coffee brand such as Starbucks or a grooming brand such as Axe. Banga focused on financial inclusion and suggested that we need to focus on the needs and behavior of consumers at the bottom of the pyramid. He further argued that governments play an important role in many social programs, and we should study the appropriate models for public/private partnerships.

4. Metrics: Almost all industry leaders voiced a need to have better metrics to help them understand consumers' multichannel behavior and the optimal way for firms to allocate resources across channels. They also expressed the need to better understand social media, to identify the best ways to use social media, and to formulate appropriate metrics to gauge their effectiveness.

5. Big data: Several practitioners expressed the need to leverage big data. Cohen highlighted the “three V's” of big data—variety, volume, and velocity. Banga thinks that combining different types of data across a large number of customers can tell us a lot about consumer behavior, and we need to find better ways to use such data more intelligently. Combining structured and unstructured data poses new challenges that need to be addressed; at the same time, we need to avoid apophenia (i.e., seeing patterns when they don't exist), argued Cohen. We note that Marketing Science announced a special issue on big data (see the call at http://pubsonline.informs.org/doi/abs/10.1287/mksc.2013.0794) and had been soliciting papers up until December 16, 2013.

Implications for Academic Research

The suggestions and ideas from the industry, if picked up by academic research, could result in high-impact findings. Moreover, they have implications for how relevant research that bridges theory and practice in marketing should be conducted. We outline some of the affected areas below.

1. Focus on substantive topics: To provide guidance on and influence how industry leaders run their businesses, we academics need to understand the issues and challenges they are facing. It was encouraging to hear that industry leaders like Lafley have high regard for academic research, but these leaders also urged us to spend our time investigating big practical problems. In other words, they urge us to focus on high-impact results (substantive or methodological) rather than make minor and incremental improvements to existing methods.

2. Managerial significance: Academic research has traditionally focused on statistical significance. However, with the availability of large amounts of data, even tiny effects are statistically significant. Faced with big data, we should thus focus on managerial substance and impact. Three-way interactions in advertising may be less crucial when the real challenge is to demonstrate a substantial main effect of a type of advertising on sales.

3. Experiments and data mining: In the digital world, it is becoming more feasible to run A/B tests to find cause and effect. If such tests can establish causality and quantify the impact of an induction, we may need to reexamine models that rely too heavily on assumptions. At the other extreme, we have a vast amount of data on social networks and social media. Data mining and machine learning have the potential to find insights in the data that would not have been possible without these methods. We need to avoid the tendency to downplay such methods as purely data driven and atheoretical. This is not to say that there is no value in theory, but these methods are helpful to identify new phenomena, new perspectives, and new theory. Once potential phenomena are uncovered, more traditional methods can be used to test and elaborate the theories. Indeed, successful scientific revolutions have always combined insights from data with new perspectives and theory. Furthermore, “normal science” means putting theories to the test—something that is increasingly feasible with big data (and methods to handle big data).

4. Real-time analysis: Traditional approaches often rely on a small sample of data. Sophisticated models might run for several days (or months) in a batch format. Although this approach is useful in many contexts, there seems to be a need for real-time iterative analysis on very large data sets. The value of a Google search inquiry depends on algorithms, databases, and computing power to return search results in a fraction of a second. Ad servers need to make decisions about which ads to serve to which consumers in real time. We are still exploring how one might sample a social network without cutting off the edges and losing the essence of the social connectivity. These challenges suggest that we need to rethink our modeling and analysis approaches.

The marketing academic community has many committed and dedicated scholars. It would be unfortunate if we fail to have a major impact on the business community and the public policy discourse. We are optimistic that the academic marketing field will pick up some of the suggestions by industry practitioners to address, in a rigorous fashion, issues that are relevant to the industry and have potentially high impact. We hope that the special section in this issue of Marketing Science is a step in this direction.