RESEARCH
Cashback is Cash Forward: Delaying a Discount to Entice Future Spending
Vana, P., Lambrecht, A., & Bertini, M. (2018). Journal of Marketing Research, 55(6), 852-868.
The authors examine purchase behavior in the context of cashback shopping—a novel form of price promotion online in which consumers initiate transactions at the website of a cashback company and, after a significant delay, receive the savings promised to them. Specifically, they analyze panel data from a large cashback company and show that, independent of the predictable effect of cashback offers on initial demand, cashback payments (1) increase the probability that consumers will make an additional purchase via the website of the cashback company and (2) increase the size of that purchase. These effects pass several robustness checks and are also meaningful: At average values in the data, an additional $1.00 in cashback payment increases the likelihood of a future transaction by .02% and spending by $.32—figures that represent 10.03% of the overall impact of a given promotion. Moreover, the authors find that consumers are more likely to spend the money returned to them at generalist retailers, such as department stores, than at other retailers. They consider three explanations for these findings; the leading hypothesis is that consumers fail to treat money as a fungible resource. They also discuss implications for cashback companies and retailers.
The Effect of Individual Online Reviews on Purchase Likelihood
Vana, P., & Lambrecht, A. (2021). Marketing Science, 40(4),708-730.
Online product reviews constitute a powerful source of information for consumers. Past research has studied the effect of aggregate measures of reviews (such as, average product rating and number of reviews) on consumer behaviour. In this study, we investigate how individual reviews displayed on a product webpage affect consumers’ purchase likelihood. Identifying this effect is challenging because retailers are free to select which reviews to display on the product page and in what order, making the display of reviews in particular positions potentially endogenous. We address this challenge by utilizing an empirical context where the retailer displays reviews by recency and exploit the variation in review positions generated as newer reviews are added on top of older ones. We find that individual reviews have a strong effect on consumer purchase decisions. These effects are particularly pronounced when individual reviews contrast with the aggregate information that is instantly available on the product page or help consumers resolve uncertainty about the product.
Complementing Human Effort in Online Reviews: A Deep Learning Approach to Automatic Content Generation and Review Synthesis
Carlson, K., Kopalle, P., Riddell, A., Rockmore, D., and Vana, P. (2023). International Journal of Research in Marketing, 40(1):54-74.
Online product reviews are one of the most ubiquitous and helpful sources of information available to consumers today for making purchase decisions. Consumers particularly rely on reviews by experts who professionally critique products for a variety of experience goods such as books, movies and wines. While these experts are capable of objectively evaluating a product, they may need assistance articulating their opinions into engaging reviews given the sheer volume of reviews they are typically tasked to write. In this paper, we seek to address this challenge by asking a broader question: “To what extent can a machine learn to write an expert review that is as engaging, informative, and appropriate as a human-written review?” We use a deep learning approach based on the Transformer network that takes as input a list of traits of the wine and generates as output a human readable text review of the wine. We apply our model to 125,000 expert reviews from Wine Enthusiast and the associated metadata including winery, style, reviewer's name, and rating. Our results suggest that the model generates reviews close to human-quality with descriptions that closely reflect the wine. In the spirit of a Turing Test, we assess through an experiment on MTurk whether the machine-generated reviews are indistinguishable to humans from human-generated reviews. We find no significant difference in respondents’ identification of whether a review was written by a machine or human being. We thus show that machines can indeed learn to write “human-quality” reviews. While extant literature focuses on using natural language processing to generate text that resembles human-written text, our main intended contribution in this research is to demonstrate that machines are indeed capable of performing the critical marketing task of writing expert reviews, work which until now has been an exclusively human task. Further, to our knowledge, there is no research on directly testing human versus machine-generated reviews. We suggest three possible applications of our model and approach and provide directions for future research.
Brands In Unsafe Places: Effects of Brand Safety Incidents on Consumers’ Brand Attitudes
Grewal, L.S., Vana, P., & Stephens, A.T. (2022). Under 4th round revision at the Journal of Marketing Research.
Well-publicized digital media incidents, in which brand content appears adjacent to “unsafe” content (e.g., negative content that is offensive, harmful, or uncomfortable), highlight the potential risk to a brand’s reputation every time it advertises on digital platforms. Even as content moderation algorithms improve, brands cannot control digital environments fully, making it imperative for marketing managers to consider brand safety processes, designed to keep a brand’s reputation safe within digital advertising ecosystems, in their risk mitigation efforts. The current research accordingly attempts to establish when brand safety concerns are more or less likely to arise, according to specific consumer-, brand-, and incident-related moderators; why consumers react negatively to incidents, depending on their capacity to erode consumer trust in brands; and how these combined elements affect various brand-related outcomes and to what extent. Across seven experiments, using both sentiment data from Twitter and stock returns, the authors distinguish brand safety incidents from other types of brand risks that demand managerial attention, and they empirically showcase how digital brand safety incidents influence consumers’ attitudes and behaviors, as well as advertisers’ outcomes. Building on these empirical findings, this article provides concrete, evidence-based suggestions for how to mitigate incidents, both before and after their occurrence.
The Impact of Algorithmic Components on Contributions in Charitable Crowdfunding
Vana, P., & Lambrecht, A. (2022). Reject and Resubmit at Journal of Marketing.
Crowdfunding platforms host thousands of projects and typically use ranking algorithms which, based on a set of project-specific variables, determine the rank order of projects in order to facilitate contributors’ choices and allow the platform to achieve specific goals. Here, we examine the role of individual components entering a ranking algorithm. We ask how such components affect project completion. We then explore whether ranking algorithms can help
direct funding towards underprivileged groups. Last, we examine the trade-offs a crowdfunding charity faces between directing funding towards underprivileged groups and having a large number of projects complete. Our study is based on data and the ranking algorithm of the educational crowdfunding platform DonorsChoose. We develop a structural model of donors’ contributions using a multiple discrete continuous choice framework and report estimation results and counterfactual outcomes if the charity reweighted algorithmic components. We find that in the algorithm the amount remaining for a project, as well as other variables related to a project’s progress, strongly affect whether a project will be fully funded. At the same time, our results indicate that prioritizing in the algorithm projects from high poverty schools increases contributions to such schools significantly. Encouragingly, our findings further suggest that, at least in our empirical context, using the algorithm to direct funding towards high poverty schools does not compromise the platform’s overall goal to collect a large amount of funding overall.
Does Amazon Have Pricing Power? Evidence from Pricing during the Covid-19 Pandemic
Vana, P., Sikdar, S., & Kadiyali, V. (2023). Reject and Resubmit at Journal of Marketing.
As a leading online retailer, Amazon is facing antitrust scrutiny about its market power. Simultaneously, Walmart is rising as an online rival. We examine daily prices of 238 products on Amazon and Walmart.com before, during and after peak Covid cases (June 2020- December 2021) for evidence of Amazon’s pricing power; Walmart prices provide an important benchmark. We find that Amazon’s key prices- the Buy Box price on which majority of transactions occur- are systematically higher than Walmart’s, and a substantial portion of these higher Buy Box prices are associated with Amazon’s third-party sellers rather than Amazon’s own seller role. However, customer ratings at Amazon are higher than at Walmart. Analysis of customer-level basket purchase data from Comscore shows Amazon customers have higher income, and hence likely higher willingness to pay. This evidence of higher prices, higher ratings and higher incomes is more consistent with market segmentation-based pricing power and does not support antitrust mitigation of market power. With its current strategy, Walmart appears to not as successfully serve its lower-income customers. We discuss implications for regulators, and managers.
How Rarity of Attributes Affects the Value of Collectible Non-Fungible Tokens: The Moderating Role of Aesthetic Evaluation of Attributes
Vana, P. (2023). Reject and Resubmit at International Journal of Research in Marketing.
This research focuses on collectible Non-Fungible Tokens (NFTs), which consist of a limited number of unique collectible images that are algorithmically generated. Each image consists of a character (typically anthropomorphic or zoomorphic) with a randomly selected set of attributes. One of the primary means through which collectible NFTs create value for their buyers is through the fact that while some attributes may appear more commonly in a large number of images, others appear rarely, so that images with rarer attributes may be more desirable. In this research, we ask how the rarity of attributes may interact with their aesthetic evaluation, a heretofore unexplored aspect of NFT design, in determining the NFT’s value. We use data and the empirical context of CryptoPunks, one of the first collectible NFT projects, which consist of 10,000 unique humanoid images. We augment the transaction data of CryptoPunks with an MTurk survey where participants evaluated each attribute of CryptoPunks along five aesthetic dimensions: familiarity, prominence, stylishness, uniqueness, and overall evaluation. We use a two-step modeling approach where the first step breaks down the sale price of each CryptoPunk as a sum of the average value of each attribute present in the image. The second step models the value of attributes obtained from the first step as a function of the attribute’s rarity, aesthetic evaluation and the interaction of rarity with aesthetic evaluation. Our results consistently show not only that the rarity of an attribute adds to its value but also that there are strong interaction effects with aesthetic evaluation. The effect sizes we find are large and managerially substantial and have relevant implications for designers of NFTs as well as buyers of NFTs.
When Crisis Hits the Platform Economy: The Effects on Supply, Demand, and Spillovers
Hong, S., Kim, J., and Vana, P. (2023). Under review at Marketing Science.
When digital platforms undergo crises due to scandals or product/service failures, the parties at stake include the two sides of “customers" that transact through the platform: the supply and the demand side. We investigate whether and how the two sides respond to a platform crisis and whether the impact spills over to a rival platform. Our empirical context focuses on the scandal surrounding Kickstarter in 2019 when it was accused of union busting. To establish the causality, we use a difference-in-differences approach using two separate control groups and show consistently that there was a decrease in the supply of projects on Kickstarter post-crisis. The drop in the probability of a project being active ranges from 5.53% to 21.02%, translating to the platform's revenue loss between $0.31 million and $1.17 million. The demand-side reaction is not as severe, showing as much or more backers’ support and pledged amounts to the active projects post-crisis. Indiegogo, Kickstarter's main rival, experienced a negative spillover on the supply side. Our analysis of creator heterogeneity informs that cultural project creators were more likely to decrease participating on Kickstarter after the crisis but less so on Indiegogo. We highlight implications for managers of platforms, supply-side agents, and policymakers.
Generating “Accurate” Online Reviews: Augmenting a Transformer-Based Approach with Structured Predictions
Vana, P., Kopalle, P., Pachigolla, P., and Carlson, K. (2024). Under review at Journal of Marketing.
A particular challenge with Generative Artificial Intelligence (GenAI) relates to the “hallucination” problem, wherein the generated content is factually incorrect. This is of particular concern for typical generative tasks in marketing. Here, we propose a two-step approach to address this issue. Our empirical context of an experience good (wines) where information about the taste of the product is important to the readers of the review but crucially, this data are unavailable a priori. Consequently, typical generative models may hallucinate this attribute in the generated review. Our approach of augmenting a transformer model with structured predictions results in a precision of .866 and a recall of .768 for the taste of wines, vastly outperforming popular benchmarks: transformer (precision .316, recall .250) and ChatGPT (precision .394, recall .243). We conduct an experimental study where respondents rated the similarity of reviews generated by our approach (versus those generated by ChatGPT) to those written by human wine experts. We find our reviews to be significantly more similar to human-expert reviews than those generated by ChatGPT. Apart from our app implementation, our main contribution in this work is to offer one approach towards more accurate GenAI, particularly towards marketing-related tasks.
Can Creative Strategy Prevent Video Ad Audience Abandonment?
Vana, P., and Neslin, S. (2024). Under review at Journal of Marketing Research.
Online video ad spend has overtaken static display ad spend. Its unique attraction – the potential to engage the customer for a nontrivial time period and get across a deeper message – ironically begets a concern. The consumer can exit out of the video, that is abandon it without watching it fully. We examine the impact of creative strategy on abandonment. Since video ads unfold over time and consumers can abandon them at any time, we study the impact of within-video variation in creative themes. We examine whether creative matters, does its impact vary over the course of the ad, and compare the impacts of informational vs. emotional creative elements. We draw on consumer behavior theory to define hypotheses related to these impacts. We adapt a customer base model that incorporates customer heterogeneity, the role of creative, and how that role changes over the course of the ad. Drawing on a database of 225 Facebook video movie trailers, we find that creative matters, and most importantly, as hypothesized, the impact of various creative themes varies over the course of the video. For example, in our application action works best at the beginning while suspense works best at the end. We also find as hypothesized that emotional and informational creative elements both influence abandonment, and that emotional creative generally is more likely to reduce abandonment than informational. We find effect sizes are meaningful – a one-standard deviation increase in a creative can influence abandonment in a range of up to 40.8%, depending on the creative element and when we increase the element. The findings suggest that managers need to carefully craft their video ads to emphasize the right creative at the right time.