Seeing Is Believing: The Impact of AI Magic Mirror on Consumer Purchase Intentions in Medical Aesthetic Services

Yu Li, Chujun Zhang, Tian Shen, Xi ChenJ. Theor. Appl. Electron. Commer. Res 2025 · DOI

TL;DR

This paper investigates how an AI-powered "Magic Mirror" tool influences consumer purchase intentions for medical aesthetic services. The study finds that the AI Magic Mirror significantly boosts purchase intentions by increasing perceived value and reducing perceived risk, especially for procedures with high "fit uncertainty" and low popularity.

What Problem Does This Paper Address?

The rapid integration of AI into e-commerce has transformed consumer experiences. While AI's role in search and experience services is well-studied, its application in "credence services" – where quality is hard to assess before purchase – remains underexplored. Medical aesthetic services are a prime example of credence services, characterized by high uncertainty, information asymmetry, and irreversible outcomes, leading to consumer hesitancy. This paper addresses this gap by examining how an AI-powered virtual preview tool, the "AI Magic Mirror," can mitigate these challenges and influence consumer decisions in this specific domain.

Methods

The study employs a multi-method approach:

  • Study 1: Secondary Data Analysis: This initial phase uses real transaction data from a Chinese medical aesthetics platform to examine the correlation between AI Magic Mirror usage and service sales. It also investigates the moderating effects of "fit uncertainty" (how well a service matches individual needs) and "procedure popularity."
  • Study 2 & 3: Experimental Studies: These studies use controlled experiments to establish a causal link between the AI Magic Mirror and purchase intentions.
  • Study 2 focuses on the main effect of the AI Magic Mirror and the moderating role of "fit uncertainty." It also tests the mediating roles of "perceived value" and "perceived risk."
    • Study 3 replicates the core findings and specifically examines the moderating role of "procedure popularity," further validating the AI Magic Mirror's impact under different conditions.

All studies utilize established scales for measuring purchase intention, perceived value, and perceived risk, adapted for the context. Statistical analyses include OLS regression, ANOVA, and PROCESS macro for mediation and moderation analysis.

Datasets

  • Study 1: Utilized a dataset of 31,386 observations from a leading Chinese medical aesthetics platform, comprising data on reviews, procedures, doctors, and hospitals.
  • Study 2 & 3: Recruited participants through online platforms. Study 2 involved 384 valid responses from an initial sample of 500, while Study 3 had 395 valid responses from an initial sample of 500. These studies used simulated scenarios of AI Magic Mirror usage versus non-usage.

Key Insights and Ideas

  • AI as a Trust-Builder in Credence Services: The AI Magic Mirror acts as a crucial tool to bridge the trust gap in medical aesthetics by allowing consumers to visualize potential outcomes, thereby reducing the inherent uncertainty of credence services.
  • Perceived Value & Risk as Mediators: The AI Magic Mirror doesn't directly impact purchase intention; it works by enhancing how consumers perceive the value of the service and diminishing their fear of risk. This highlights AI's role in reframing consumer evaluation.
  • Fit Uncertainty Amplifies AI's Impact: For procedures where individual fit is highly variable and uncertain (e.g., double eyelid surgery), the AI Magic Mirror is significantly more effective in boosting purchase intentions. It helps users visualize personalized results, addressing their specific concerns.
  • Popularity Dampens AI's Effect: Conversely, for highly popular procedures, consumers already have ample information and reduced uncertainty, making the AI Magic Mirror less critical. Its impact is more pronounced for less familiar or less popular services.

Key Results

  • Main Effect: The AI Magic Mirror consistently and significantly increases consumer purchase intentions for medical aesthetic services across all studies. Study 2 showed a 17% increase in purchase intentions.
  • Mediation: Perceived value significantly mediates the positive relationship between AI Magic Mirror use and purchase intention. Perceived risk also mediates, but negatively (AI Magic Mirror reduces risk, which increases purchase intention).
  • Moderation (Fit Uncertainty): Higher fit uncertainty strengthens the positive impact of the AI Magic Mirror on purchase intentions.
  • Moderation (Procedure Popularity): Lower procedure popularity strengthens the positive impact of the AI Magic Mirror on purchase intentions. The effect was significantly weaker for high-popularity procedures.

Limitations and Future Work

  • Study 1 Limitations: The secondary data was incomplete, and the analysis was correlational, preventing definitive causal claims.
  • Generalizability: The studies primarily used samples from China, and cultural differences might influence consumer responses to AI. Future research could explore cross-cultural variations.
  • Psychological Factors: The study focused on perceived value and risk, but future work could explore deeper psychological factors like appearance anxiety, self-identity, and social comparison pressure.
  • Boundary Conditions: While fit uncertainty and popularity were examined, future research could investigate other boundary conditions, such as hospital or physician attributes.
  • Real-World vs. Simulated: Experimental studies, while controlled, may not fully replicate real-world service contexts. Future research could involve natural experiments in collaboration with platforms.

Summary updated

BibTeX
@article{li2025seeing,
  author = {Yu Li and Chujun Zhang and Tian Shen and Xi Chen},
  title = {{Seeing Is Believing: The Impact of AI Magic Mirror on Consumer Purchase Intentions in Medical Aesthetic Services}},
  journal = {J. Theor. Appl. Electron. Commer. Res.},
  year = {2025},
  doi = {10.3390/jtaer20030205},
}

Loading related content...

Figure