Interpretable Recommendations and Parameter-Grounded Explanations with Multi-Graph Attention
As AI-assisted search and service interactions become ubiquitous, recommender systems must demonstrate the transparency needed to persuade customers and earn their trust.  This approach shows how to engineer such systems to deliver interpretable predictions and user-centered explanations that are grounded in the same quality and social signals used for prediction.
Author: Yan Leng, Xiao Liu and Rodrigo Ruiz

Designing Consent: Choice Architecture and Consumer Welfare in Data Sharing
Privacy regulations require consent to use tracking “cookies,” but how consumers respond depends on the choice architecture of notifications. Most currently “accept all” or incorrectly assume that dismissing a notice rejects all. Interfaces that clearly display all options or a universal browser setting that applies across websites maximize consumer welfare.
Author: Tesary Lin

Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia
Search engine AI overviews reallocate attention away from publisher pages, reducing advertising, subscription and sales revenue. Publishers should strengthen direct channels and optimize for citation with content less vulnerable to substitution. Sustaining the content ecosystem on which AI summaries depend may require stronger attribution, more meaningful referral mechanisms, and revenue-sharing
Author: Mehrzad Khosravi and Hema Yoganarasimhan

From Reviews to Responses: Bridging Pre- and Post-Purchase Consumers through AI-Enhanced QA with RAG
Websites struggle to combine consumer reviews and question-answer systems that influence shopper purchase decisions. Firms should incentivize users to contribute both reviews and questions. Platforms can then model this proprietary data with Retrieval Augmented Generation LLMs, a question–review matching module, and an answerability classifier to enhance the pre-purchase experience.
Author: XingJian You, Ishita Chakraborty and Neeraj Arora

Why Segmented Food Portions Appear Larger: Evidence for Numerosity and Surface Area but Not Reassembly Difficulty, with Implications for Smarter Downsizing
Controlling the size of food portions can reduce overconsumption and waste, but size is in the eye of the beholder. Increasing the number of portions and the separation among them increases total perceived size. Such segmentation can reinforce social norms about appropriate portion size and increase consumer willingness to pay.
Author: Pierre Chandon, Andde Indaburu and Maria Langlois

Targeting Information in Ad Auction Mechanisms
Advertisers bid for online ads based on demographics and other data provided by publishers. Granular data can improve ad matching but also reduce publishers’ revenues by reducing bids for specific inventory. Both parties can benefit from auctions with mixed bundles of inventory types that broaden competition but also improve targeting.
Author: Srinivas Tunuguntla, Carl F. Mela and Jason Pratt

Open-Source Media and Marketing Mix Modeling: Overview, Challenges, Opportunities
Automated open-source media and marketing mix models can help resource-constrained firms measure incremental marketing impact across channels but may also introduce managerial risks by concealing model misspecifications, insufficient variation and endogeneity. Calibration and triangulation using experiments, trusted attribution signals and structured diagnostics can keep observational estimates tethered to business reality.
Author: Julian Runge, Koen Pauwels

Learning from Many Experiments: A Hierarchical Bayesian Framework for Decomposing Marketing Treatment Heterogeneity
A Bayesian model integrating multiple marketing interventions over time combines the causal-inference advantages of randomized experimentation with the longitudinal richness of repeated observational data. Results reveal that marketing effectiveness depends much more on customer receptivity than campaign design or timing, pointing to the need for—and means of—better targeting.
Author: Peter Ebbes, Eva Ascarza, Oded Netzer

The Impact of LLM Adoption on Online User Behavior
Early LLMs substituted for traditional search engines, reducing website traffic and ad exposure by 20 percent. Advanced LLMs will accelerate this competition. Publishers must offer high-value, differentiated content to attract traffic. LLMs and search engines need to compensate and incentivize those who create the content that maintains the information ecosystem.
Author: Nicolas Padilla, Tai Lam, Anja Lambrecht, Brett Hollenbeck

Content Creator or Self-Journaler? Classifying Social Media Influencers in the Creator Economy
Social influencers vary by their relative focus on content versus self-presentation. Self-journalers attract larger audiences for broader reach across categories. Content creators deliver expertise and engagement intensity for more focused followers. Self-creators bridge the difference. Brands can benefit from combining all three to address different stages of the customer journey.
Author: Gloria Peggiani, Lucio Lamberti and Koen Pauwels

Worthy of Your Binge: How Media Momentum Drives Satisfaction in Clumped Consumption
As “binge watching” of media increases, how can platforms and audiences avoid the hangover? Shows combining complexity and emotional engagement have “media momentum” that drives viewing and satisfaction. Complexity counteracts hedonic adaptation, keeping the consumer involved, and resonance keeps the experience pleasant despite the effort required to follow the plot.
Author: Rachele Ciulli, Cait Lamberton and Robert Meyer

Predicted Incrementality by Experimentation (PIE) for Ad Measurement
Random controlled trials (RCTs) can assess causal effects of marketing but are expensive and incur opportunity costs by excluding control groups. Predicted Incrementality by Experimentation (PIE) uses samples of RCT-run campaigns to determine which characteristics map to causal outcomes and then applies that mapping to campaigns not run as RCTs.
Author: Brett R. Gordon, Robert Moakler, Florian Zettelmeyer

Selling the Haggle
Common in personal selling markets, price negotiation can be seen as “friction” for both customers and companies—but fixed pricing is not necessarily the answer. Haggling is a sales tool that often serves to close the deal. Removing it risks sales losses that offset the intended gains from price transparency.
Author: Keyan Li, Zelin Li, Siqi Pei and Feng Yang

Generative AI and Firm Productivity: Field Experiments in Online Retail
GenAI may enhance an individual’s productivity, but can this scale to justify corporate investments? In online retail, holding implementation and organizational factors constant, GenAI assisted workflows can increase revenue by reducing friction at several touchpoints across the customer journey. Less experienced consumers and smaller and newer sellers benefit the most.
Author: Lu Fang, Zhe Yuan, Kaifu Zhang, Dante Donati and Miklos Sarvary

Shopping with a Platform AI Assistant: Who Adopts, When in the Journey, and What For?
As online retailers rapidly implement AI shopping assistants, they may be aiming at the wrong users. Unlike the younger, male, tech-savvy stereotype, early adopters of AI assistants tend to be older, female and those already most engaged with and loyal to the platform. AI complements rather than replaces conventional search.
Author: Se Yan, Han Zhong, Zemin (Zachary) Zhong and Wenyu Zhou

TextBO: Bayesian Optimization in Language Space for Eval-Efficient Self-Improving AI
Develops and validates an iterative prompt-optimization-based self-improving AI that uses Bayesian Optimization in the language space to generate prompts for high-performing ads using the fewest evaluations possible.
Author: Enoch H. Kang and Hema Yoganarasimhan