| 摘要: | COVID-19 疫情加速了數位消費成長，客服需求激增，企業廣泛採用客服機器人處理消費者的問題與申訴，以提升服務效率並降低營運成本。然而，現有客服機器人雖能模擬人類的應答方式，卻仍難以提供情感支持與真誠互動，影響服務補救成效；情緒智商的展現為提升互動品質的關鍵之一。
本研究以刺激－個體生理或心理－反應（Stimulus－Organism－Response, SOR）理論為基礎，延伸建構「溝通策略－感知與認知－反應」（Communication Strategy－Perception－Response, CPR）模型，旨在探討服務失敗情境中，客服機器人採用不同溝
通策略對消費者負面情緒、互動正義感知、誠意感知與後續行為意圖之影響，並檢驗情緒智商與社會距離的調節作用。
研究結果顯示，相較於中立回應，情感支持結合幽默的溝通策略在服務補救中更能有效緩解消費者情緒與提升消費者對整體服務的評價；其中良性回應對情緒緩解成效最佳，自我懲罰回應則最能提升互動正義與誠意感知。調節效果方面，情緒智商僅於中立回應中具顯著作用，顯示在缺乏情感線索的互動中，個體特質成為評價誠意的重要因素；社會距離則僅於良性回應組中具顯著負向的調節作用，反映親和性語氣有助於縮短心理距離，進一步提升服務補救滿意度。
本研究填補客服機器人補救人類服務失誤、幽默策略應用於外送平台場景之研究缺口，並提供階段式策略設計等實務建議，對智慧客服系統之人性化互動設計具重要參考價值。;The COVID-19 pandemic has accelerated the growth of digital consumption, leading to a surge in customer service demands. In response, many companies have adopted customer service chatbots to handle consumer inquiries and complaints, aiming to enhance service efficiency and reduce operational costs. However, despite their capacity to simulate humanlike responses, these chatbots often struggle to provide genuine emotional support and sincerity—key factors influencing the effectiveness of service recovery. The demonstration of emotional intelligence is considered a critical element in improving the quality of human–AI interaction.
Grounded in the Stimulus–Organism–Response (SOR) theory, this study extends the framework by developing a Communication Strategy–Perception–Response (CPR) model. It examines how different communication strategies employed by chatbots influence consumers’negative emotions, perceptions of interactional justice, perceived sincerity, and subsequent behavioral intentions in service failure contexts. Furthermore, it explores the moderating roles of emotional intelligence and perceived social distance.
The results reveal that, compared to neutral responses, communication strategies combining emotional support and humor more effectively alleviate customer negative emotions and enhance their overall evaluation of service recovery. Among these strategies, affiliative humor proves most effective in relieving negative emotions, while self-defeating
humor most significantly enhances perceived interactional justice and sincerity. Moderation analyses indicate that emotional intelligence exerts a significant moderating effect only in the neutral response condition, suggesting that individual traits are particularly influential when emotional cues are absent. Perceived social distance exhibits a significant negative moderating effect only in the affiliative humor condition, highlighting that socially close, empathetic language helps reduce psychological distance and improves service recovery satisfaction.
This study contributes to the literature by addressing the underexplored context of chatbot-mediated recovery for human service failures and by applying humor-based strategies in food delivery platforms. It offers practical implications for staged strategy design and humanized response development, providing valuable insights for the design of emotionally intelligent customer service chatbots. |
