2025 MSI Summit: Presentations, Recordings, and Summaries
Marketing Analytics in the Age of AI
Jean-Pierre Dubé – James M. Kilts Distinguished Service Professor of Marketing, University of Chicago Booth School of Business
What are the misconceptions about AI, its literal meaning, and plausible expectations for how it will impact marketing in the near term? Learn from two real-world case studies to demonstrate how the AI toolkit can help companies make better data-based marketing decisions.
Synergizing Past Experiments for Intervention Personalization
Eva Ascarza – Jakurski Family Associate Professor of Business Administration, Marketing Unit, Harvard Business School, Harvard University
This research presents Incrementality Representation Learning (IRL), an advanced machine learning framework designed to help businesses harness past marketing experiments to predict the impact of new interventions, eliminating the need for costly and extensive testing. By integrating customer data with intervention design features, IRL enhances targeting and personalization, empowering firms to craft tailored marketing strategies that drive both profitability and customer engagement. Validated through large-scale CPG campaigns, IRL significantly outperforms traditional methods in predicting treatment effects and optimizing interventions, delivering superior results for new customer segments and untested promotional offers.
Research Priority Workshop Framework
Keith Smith, Ph.D. – MSI Managing Director
MSI’s Research Priorities provide a framework for future academic research and industry-focused programming. In this session, we will quickly review our current priorities and outline the work of the Research Priority breakouts.
AI-Powered Insights for a Cookie-Free Marketing Future: Modeling Multi-Touch Consumer Journeys
Alice Li – Associate Professor of Marketing, Department of Marketing & Logistics, The Ohio State University
Firms face growing uncertainty as cookies phase out, prompting various hypotheses and contingency plans. My co-authors and I are developing models to simulate firm responses to this shift and explore effective transition strategies. Additionally, we leverage aggregate time-series data for forecasting and bridge multi-touch attribution with marketing mix models. These methods provide actionable, data-driven solutions to help firms navigate a cookie-free future.
The Transparency Dilemma: How AI Disclosure Erodes Customer Trust and What Firms Can Do About It
Martin Reimann – McClelland Associate Professor of Marketing, Eller College of Management, University of Arizona
As generative artificial intelligence tools like ChatGPT are becoming an integral part of an increasing number of business processes, marketing executives face important decisions about whether to proactively disclose their use of AI to marketplace stakeholders. Does disclosing AI usage undermine trust in the actor using it? Hear about vital insights for marketing leaders navigating the balance between transparency and trust in AI-powered marketing practices. While transparency is usually viewed as a virtue, disclosing AI usage carries unintended consequences.
Consumers’ Perception of AI-Generated Marketing
Fabian Buder – Head of Future & Trends Research, Nuremburg Institute for Market Decisions
This presentation explores global consumer attitudes toward AI-generated marketing content, revealing widespread skepticism and a preference for human-created alternatives. Drawing on a series of experiments and representative cross-cultural survey data, it highlights key barriers and will discuss steps for businesses to enhance acceptance and trust in marketing and advertising.
Visual Drivers of Emotions: A Model for Extracting Emotional Loading of Marketing Images
Natalie Mizik – Professor of Marketing and J. Gary Shansby Endowed Chair in Marketing Strategy, University of Washington Foster School of Business
We use computer vision and deep learning to extract predictors of emotion elicited by marketing images. We consider: (1) elements of design (low-level visual features) such as color, texture, shape, lines, curves, corners, edges, and orientation, (2) high-level objects (e.g., adventure, action, leisure, danger, etc.), (3) human facial expressions, and (4) text embedded in the image and train XGboost models to predict image Sentiment and Arousal.
Using AI to Derive Causal Relationships from Qualitative Data
Kate Weymer – Director Data Science, Microsoft
An overview of how AI in the form of LLMs and causal inference algorithms bring scaled precision and actionability to our qualitative data.
Theory-Aligned Knowledge Distillation with LLMs: Applications to Content Marketing and Customer Service
K. Sudhir – James Frank Professor of Marketing, Private Enterprise, and Management, Yale School of Management
Learn about two applications of large language models (LLMs) in content marketing and customer service, focusing on how the knowledge embedded in LLMs can be leveraged to optimize performance in each context with interpretable theory-based guidance. The content marketing example demonstrates how LLMs can integrate A/B test results to generate and validate theories for optimizing headlines, while the customer service application shows how knowledge distillation from advanced LLMs can enhance the performance of low-cost, open-source models.