May 29–30, 2025
Check-in at Hanover Inn begins at 3:00 p.m. Rooms are reserved under your name.
7:00 p.m.
Dinner at Pine, in the Hanover Inn.
8:15–8:45 a.m.
Continental Breakfast in Alperin at the Tuck School of Business
8:45–9:00 a.m.
Welcome by Praveen Kopalle Signal Companies’ Professor of Management & Professor of Marketing, Tuck School of Business
9:00–10:00 a.m.
Lisa Cavanaugh, Associate Professor of Marketing and Behavioural Science, University of British Columbia
“When Plus-size Models Help Versus Hinder Brand Outcomes
Break
10:15–11:15 a.m.
Shane Wang, Professor of Marketing, Pamplin College of Business, Virginia Tech
“Predicting Consumer Behaviors with Large Language Model (LLM)-Powered Digital Twins of Customers”
Break
11:30–12:30 p.m.
Maura Scott, Professor and Edward M. Carson, Chair in Services Marketing, W.P. Carey School of Business, Arizona State University
“The Role of Marketing in Developing Services to Promote Individual and Societal Well-Being”
12:30–1:30 p.m.
Lunch in Alperin
1:30–2:30 p.m.
Ali Goli, Assistant Professor of Marketing, Foster School of Business, University of Washington
“Retail Pricing and Organizational Structure”
Break
2:45–3:45 p.m.
Sharmistha Sikdar, Assistant Professor of Business Administration, Tuck School of Business
“Managing the Customer Journey to Generate Purchase at Maximum Profitability”
3:45–6:00 p.m.
Free Time
7:00 p.m.
Dinner at Murphy’s on the Green, Hanover
Visit concludes. Check-out time at the Hanover Inn is 11:00 a.m.
Title: When Plus-size Models Help Versus Hinder Brand Outcomes
Abstract: In recent years, brands have dedicated increased attention to body-inclusivity in marketing. Responding to expectations for representation and expressed consumer demand, brands have employed a wider range of model sizes in marketing their products. An important question is whether and how this practice impacts brand outcomes. This research reveals when marketing products by using a mix of model body sizes can benefit and backfire for brands. This research shows that using a mix of slender and plus-size models (vs. only slender models) helps brands by enhancing consumers’ brand attitudes but hinders brands by causing swings in consumers’ product preferences and choices within a product assortment. Experiments and a field study document an attitude-action gap in consumer responses to body-inclusive marketing whereby consumers’ brand attitudes and product choices often diverge. These findings have important managerial implications for both product marketing and brand management.
Title: Predicting Consumer Behaviors with Large Language Model (LLM)-Powered Digital Twins of Customers
Abstract: We propose and test a methodological framework combining retrieval-augmented generation and fine-tuning of large language models, with user-generated content, to implement digital twins of customers (DToC). They have been touted as a promising marketing tool to understand and predict consumer preferences and behaviors. The digital twins leverage existing consumer profiles and behavioral data; by building virtual models of consumer(s) that emulate their way of thinking, feeling, and decision-making, marketers are able to predict the product consumers purchase next, test the effectiveness of targeted marketing communications, improve customer experience along the customer journey, and so on. Now, with the advancement of Generative AI and large language models, digital twins of customers are evolving into living simulacra that can be connected to real-time consumer data and fine-tuned to learn, simulate, and predict consumer behaviors. More importantly, they can provide dynamic responses and directly interact with marketers for different marketing purposes. This research demonstrates the efficacy of this approach through the empirical applications of predicting online shopping behaviors. Compared to existing LLM-based methods relying largely on prompt engineering, our framework fine-tunes models for greater context awareness and real-time adaptability, achieving improved accuracy in behavior prediction. Practically, the DToC framework enables marketers to simulate customer responses to campaigns, products, and offerings without direct engagement, reducing costs while improving marketing decision-making.
Title: The Role of Marketing in Developing Services to Promote Individual and Societal Well-Being
Abstract: This presentation discusses insights from ongoing projects that examine barriers to consumers’ well-being and services designed to promote individual and societal well-being.
Project 1: This research examines the journeys of refugees, including how integration programs differentially influence women and men in terms of their well-being. Specifically, in this multi- method research, we uncover how integration training programs help to differentially reduce fears among refugees as a function of their gender. In Study 1, we examine the experiences of over 6000 refugees. We find that female refugees experience greater levels of fears, and it takes longer to mitigate their fears as they strive for integration in their new country. In Study 2, we conduct qualitative interviews with 19 refugee women. We find that several factors constrain their wellbeing including cultural factors (e.g., roles of women in the culture vs. the new country), time constraints, and social factors (loss of identity and discrimination experiences). In Study 3, we conduct interviews with service providers focused on refugee integration, in which we discuss the challenges unique to refugee women and potential interventions. In studies 4 and 5, we experimentally test interventions that help to create a more positive integration experience and increase wellbeing among refugee women.
Project 2: This research, in collaboration with a financial services firm, examines how to help increase household financial stability and wealth. In Study 1, we find that consumers are more interested in financial assistance programs that focus on savings increases (relative to debt reduction). Study 2, a field study in collaboration with a financial institution, analyzes a program designed to help consumers improve their financial well-being. We find that, over time, customers in the financial well-being program are able to reduce their debt levels to a greater degree than customers not participating in the program; however, they do not show differences in increasing their savings. Study 3, a field study with the firm’s customers, examines interventions to encourage consumers to increase savings in both the short-term (e.g., emergency funds) and long-term (e.g., retirement).
Title: Retail Pricing and Organizational Structure
Abstract: We document the central role of organizational structure in determining pricing strategies in the U.S. food retail industry. Large parent companies that own multiple chains employ zone pricing aligned with their organizational divisions, reflecting delegation of pricing authority from central to divisional management. While chains typically belong to a single zone, explaining the uniform chain-level pricing commonly found in previous studies, parent-level pricing is more complex. Multiple chains can operate under coordinated pricing within a single zone, and some chains can span multiple zones with different prices. Price zones generally cover separate geographic areas, and in cases when the zones do overlap they coincide with “discounter” and “traditional” chains operating at different price points. Price zones are deeply rooted in organizational divisions and costly to adjust, as evidenced by a two-year restructuring effort following a large merger. Yet, parent companies can exhibit flexibility when required: the enactment of soda taxes demonstrates their ability to implement granular within-zone price adjustments in response to local economic shocks.
Title: Managing the Customer Journey to Generate Purchase at Maximum Profitability
Abstract: In online retailing, customer journeys often begin with firm-driven marketing or self- directed searches, progressing through visits that may lead to a purchase. Firms aim to optimize communication strategies to encourage visits and sales, that require careful decisions on when, how, and whom to target. For instance, a recent visitor may not need additional communication, while a lapsed visitor might benefit from a timely email. However, misdirected communication can waste resources or deter customers.
This is especially challenging for durable goods or experiences, such as vacation packages, where purchases are infrequent, and inactivity may signal journey abandonment. To address this, firms need models to predict customer behavior and optimize communication strategies. This research develops a two-part optimal contact model: (1) a latent class multinomial (LC-MNL) model that predicts customer visits, purchases, and inactivity based on past behavior and communications, and 2) a reinforcement learning or Q-learning algorithm that uses this predictive model to determine optimal marketing interventions.
We identify four latent classes based on our LC-MNL model estimates with the segments varying in terms of their baseline visit-purchase behavior, as well as that as a function of marketing actions. The model-based Q-learning optimization determines the optimal targeting policy for the agent (i.e., firm).
Our results merit several insights into how marketing actions vary in frequency and type across the segments. We find that low performing segments may merit more investment. This is consistent with recommendations from extant literature on the “recency trap” and why firms should not ignore low frequency customers.