Is Customer Service Innovation Always Preferable? The Impact of AI Technology on an Online Retailer’s Customer Service Decision

Leilei Zhao, Weiwei WuJournal of Theoretical and Applied Electronic Commerce Research 2025 · DOI

TL;DR

This paper uses a theoretical model to explore whether online retailers should opt for human or AI-powered customer service. It reveals that while AI can be more efficient, it doesn't always lead to better outcomes for retailers, as consumer preference and the cost of service play crucial roles.

What Problem Does This Paper Address?

In today's competitive online retail landscape, customer service is a critical differentiator. As AI technology advances, online retailers are increasingly adopting AI-powered chatbots and assistants for customer service. However, there's a prevailing assumption that innovation in customer service, particularly the adoption of AI, is always beneficial. This paper questions this assumption by investigating the strategic decision-making of online retailers when choosing between human customer service and AI customer service. It aims to understand the conditions under which each strategy is more profitable and preferred by consumers, considering factors like problem resolution rates, consumer acceptance of AI, and consulting costs.

Methods

The authors employ a stylized analytical model to represent an online retailer selling directly to consumers. They consider two distinct customer service strategies:

  • Model H (Human Customer Service): Retailers hire human employees to handle customer inquiries.
  • Model A (AI Customer Service): Retailers utilize AI technology (like chatbots) for customer service.

The model incorporates key variables such as:

  • Consumer valuation (v): The inherent value a consumer places on a product.
  • Product price (p): The price set by the retailer.
  • Service level (e): The quality or effectiveness of the customer service.
  • Problem resolution probability (θ): The likelihood that a customer's issue is resolved.
  • Consumer acceptance of AI (β): The degree to which consumers are willing to interact with AI.
  • Consulting cost (l): The cost (time, effort) for consumers to use customer service.
  • Retailer's cost coefficient (k): The cost incurred by the retailer for providing service.

By analyzing the retailer's profit function under each model and comparing the equilibrium outcomes, the study identifies the factors that influence the retailer's choice and the resulting market dynamics.

Datasets

Not discussed in the paper. The study relies on a theoretical model and analytical derivations rather than empirical data for training or evaluation.

Key Insights and Ideas

  • AI isn't always better: Contrary to the intuitive belief that advanced AI is always superior, this study shows that AI customer service with higher performance doesn't automatically guarantee better outcomes for the retailer.
  • Consumer preference is nuanced: While existing research suggests consumers prefer human service, this paper finds that consumers can prefer AI service in specific situations, particularly when consulting costs are high and AI is effective at resolving problems.
  • Cost and resolution are key drivers: The probability of resolving a consumer's problem and the consulting cost significantly influence the retailer's decision between human and AI customer service.
  • Consumer acceptance of AI matters: The degree to which consumers accept AI technology plays a critical role in determining the profitability of AI customer service. Low acceptance can negate the benefits of AI's efficiency.
  • U-shaped relationships exist: The optimal price, service level, and profit for retailers can exhibit a U-shaped relationship with the number of consumers experiencing shopping problems, depending on the customer service strategy and AI acceptance.

Key Results

The paper presents several propositions and corollaries detailing the impact of various factors on pricing, service levels, and retailer profits:

  • Human Service (Model H): Optimal price, service level, and profit can have a U-shaped relationship with the number of consumers with shopping problems. These metrics generally increase with the probability of problem resolution (θH).
  • AI Service (Model A): The impact of consumers with shopping problems on outcomes is heavily influenced by consumer acceptance of AI (β). High acceptance leads to positive relationships, while low acceptance can lead to decreased prices, service levels, and profits. Optimal price, service level, and profit increase with the probability of problem resolution (θA).
  • Comparative Analysis:
    • The comparison of product prices and service levels between human and AI service depends on the interplay between problem resolution probabilities (θH vs. θA) and consulting costs (l).
  • Retailer preference for one strategy over the other is determined by a complex combination of these factors, including consumer acceptance of AI. For instance, when AI's problem resolution probability is higher, retailers might favor AI, especially if consulting costs are high, even if consumers have a slight bias towards human service. Conversely, if consulting costs are low, consumers might lean towards human service, influencing the retailer's decision.

Limitations and Future Work

Limitations: * The study uses a stylized theoretical model, which simplifies real-world complexities. * The analysis focuses on a single online retailer and doesn't explicitly model competition between multiple retailers using different strategies. * Consumer segmentation is simplified, with only two types of consumers considered.

Future Work: * Investigate consumer preferences for AI and human customer services in competitive environments with multiple retailers. * Explore the impact of AI customer service on retailers' channel strategies and competition in the context of advertising. * Examine the influence of different types of online customer services on product return issues.

Summary updated

Abstract

The online shopping consulting service (online customer service) is an important component of online shopping, and the rapid advancement of artificial intelligence technology has allowed online retailers to apply this technology to customer service. This study explores an online retailer’s customer service innovation decisions. We develop a stylized model to investigate an online retailer’s choices of two different online shopping consulting service strategies, as well as the impact of the different strategies on both online retailer and consumers. This study demonstrates that providing AI customer service with higher performance and higher probability of consumer shopping problem resolution may not always benefit the online retailer. Existing studies have already shown that consumers prefer human customer service. However, this study shows that consumers do not always prefer human customer service, and they prefer AI customer service in some situations. Moreover, this study reveals that the probability of consumers’ shopping problems being resolved and the consulting cost can affect the online retailer’s choice of online customer service strategies.
BibTeX
@article{zhao2025is,
  author = {Leilei Zhao and Weiwei Wu},
  title = {{Is Customer Service Innovation Always Preferable? The Impact of AI Technology on an Online Retailer’s Customer Service Decision}},
  journal = {Journal of Theoretical and Applied Electronic Commerce Research},
  year = {2025},
  doi = {10.3390/jtaer20020101},
}
Figure