Published work

Cross-Platform Spillover Effects in Consumption of Viral Content: A Quasi-Experimental Analysis Using Synthetic Controls

Haris Krijestorac, Rajiv Garg, and Vijay Mahajan

Published in Information Systems Research (2020)

Decisions Under the Illusion of Objectivity: Digital Embeddedness and B2B Purchasing

Haris Krijestorac, Rajiv Garg, and Prabhudev Konana

Published in Production and Operations Management (2021)

Working Papers

Personality-Based Content Engineering for Rich Digital Media

Haris Krijestorac, Rajiv Garg, and Maytal Saar-Tsechansky

In Progress

Voice of the Machine: The Role of Vocal Acoustics in Consumer Decision-Making

Haris Krijestorac, Rajiv Garg, and Vijay Mahajan


With the rise of speech-driven digital interfaces, such as Alexa and Siri, voice is increasingly utilized to interact with consumers. While such interfaces typically offer a variety of synthetic voices for users to choose from, there is little evidence to inform the design of these voices, or to tailor them to users or context. To uncover such insight, we analyze voices based on a comprehensive set of acoustic properties – such as pitch, volume, and tempo – and examine their effectiveness in a marketing context. To achieve this, we first empirically identify and label six generalizable voice clusters, using acoustic features extracted from 571 voices. We then present 355 subjects with AI-generated voice ads for hedonic and utilitarian products. Based on a survey of reactions to these ads, we find that “ostentatious” voices tend to be more effective, while “seductive” ones are less effective. Meanwhile, “suave” and “colloquial” voices are more effective for hedonic products only. We find that female voices are better at encouraging information seeking, while male consumers are more responsive to voice ads. We discuss implications for voice-enabled devices and voice actors, as well as future directions in voice analytics.

Arcs of Triumph: Fostering User Engagement with Social Media Videos through Emotional Arcs

Reza Alibakhshi, Haris Krijestorac, and Shirish Srivastava


As short-form videos on social media platforms (e.g., Instagram, Snapchat, TikTok) gain traction, a challenge for content creators is converting passive consumption of these media (views), which is plentiful, into active engagement (likes), which is comparatively rare. Prior studies demonstrate that narrative arcs, whether in the context of movies or sports, are key to stimulating such engagement with media. Building on this, our study explores how embedding strategic “emotional arcs” (i.e., the expression of various emotions over time) in Instagram videos may convert “views” into “likes”. Using a dataset of 765 videos from 80 corporate Instagram accounts, we employ machine learning tools to extract levels of expression of five key emotions over time within these videos. Leveraging an ensemble of empirical approaches, we demonstrate that emotional arcs alone can predict with 81% accuracy whether a video is high- or low-engagement, suggesting that these arcs are indeed meaningful to stimulating user interaction. Building on this, we unpack the black box to understand what specific emotional arcs lead to high engagement, and support these findings through causal analysis. Overall, our results suggest that considering multiple complex emotions, including a mixture of positive (e.g., happiness), and negative (e.g., anger, fear) ones, is most effective at fostering engagement. Considering the arcs of such emotions, our results suggest that their moment-to-moment predictability, rather than their variability, is key to engagement. We discuss implications for content creators, marketers, and platforms.