How Virtual Celebrity Characteristics Drive Purchase Intention: Testing the Stimulus-Organism-Response Framework with Structural Equation Modeling

Summary

In the context of the virtual celebrity experience, content quality most strongly drives satisfaction; satisfaction, in turn, increases purchase intention indirectly via loyalty. Personalization strengthens the satisfaction-loyalty link, so firms should prioritize emotion-evoking, narrative-consistent virtual celebrity content and personalized loyalty programs.

Abstract

In recent years, firms have begun introducing virtual celebrities to promote their products and brands across various media channels. This research employs the stimulus-organism-response (SOR) framework to empirically investigate how virtual celebrity attractiveness factors, such as appearance, voice, and content quality, can influence consumer satisfaction, loyalty, and purchase intention. Partial least squares structural equation modeling (PLS-SEM) was used to test the hypotheses based on a sample of 249 participants who had engaged in virtual celebrity experiences. The findings indicate that content quality is the strongest predictor of user satisfaction and that satisfaction indirectly influences purchase intention through loyalty. Furthermore, personalization moderates the satisfaction-loyalty relationship, suggesting that personalized experiences would enhance the engagement of virtual celebrities and fans. The findings indicate that firms considering virtual celebrity campaigns should focus more on emotion-eliciting, narrative-consistent content rather than visual realism. Furthermore, investments in loyalty-building programs and personalized engagement solutions are necessary to transform consumer satisfaction into sustained behaviors.

Introduction

Celebrity endorsement remains widely used, and firms increasingly adopt virtual celebrities across digital media channels¹. Advances in AI and related technologies have diversified virtual celebrity forms and improved scalability and interaction².

Virtual celebrities are hybrids of virtual characters and traditional celebrities³ and are defined here as digitally created personas that achieve public recognition and sustained attachment through recurring narrative-based interaction^3,4. This study focuses on audiov....

Protocol

The study qualified for exemption from full ethical review under institutional minimal-risk guidelines, as it involved an anonymous online survey with no personally identifiable information collected. An informed consent form describing the study purpose, voluntary participation, anonymity, and data usage was presented at the beginning of the survey, and only consenting participants proceeded.

1. Sample size determination

The final sample (N = 249) was determined based on statistical power requirements for PLS–SEM. The maximum number of predictors for a single endogenous construct was six (th....

Results

Table 2 summarizes the demographic characteristics of the final sample (N = 249), including gender, age, education level, and frequency of engagement with virtual celebrities. The sample showed balanced gender representation (52.6% female) and consisted primarily of young to middle-aged adults with relatively high educational attainment. Engagement frequency was concentrated in the weekly–monthly range, with fewer users per day. When included as a control variable, usage frequency did not significantly a.......

Discussion

This study examines how virtual celebrity characteristics influence satisfaction, loyalty, and purchase intention within an SOR framework, with personalization as a moderator. Three key insights emerge: content quality is the strongest driver of satisfaction relative to appearance and voice; satisfaction influences purchase intention only indirectly through loyalty (indirect-only mediation); and personalization strengthens the satisfaction–loyalty relationship but does not directly affect purchase intention.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Qualtrics Version: N/A	Qualtrics, LLC	N/A	Survey coding, https://www.qualtrics.com/
Prolific Version: N/A	Prolific	N/A	Online survey pannel provider, https://www.prolific.com/
SmartPLS Version: 4.0.1	SmartPLS GmbH	N/A	PLS-SEM analysis,https://www.smartpls.com/
ChatGPT Version: 5.2	OpenAI	N/A	Enhancing readability and grammar check, https://www.chatgpt.com/