Subscription retention is crucial for a firm's sustained success, with the expectation-confirmation model (ECM) commonly used to explain users' intentions to maintain subscriptions. However, ECM shows limitations when applied to ChatGPT Plus, as users may continue their subscriptions even amidst fluctuating satisfaction. ChatGPT, a well-known artificial intelligence (AI) tool, frequently generates incorrect or misleading responses, yet users often retain their subscriptions. This study seeks to address gaps in ECM by empirically investigating the mediating role of goal-congruent outcomes to better identify the factors influencing users' subscription continuance intentions.
A survey was conducted with users who have subscribed to ChatGPT Plus for over three months. The dataset comprises 333 valid responses from an online survey conducted via the Prolific platform. IBM SPSS and AMOS 31 were used to analyze the proposed model and the relationships among variables.
The results show that satisfaction has a strong and significant effect on continuance intention, while confirmation positively influences both perceived usefulness and satisfaction. In contrast, perceived usefulness does not directly predict satisfaction or continuance intention. Instead, goal-congruent outcomes fully mediate the effects of perceived usefulness on both satisfaction and continuance intention, highlighting their central role in sustaining long-term subscription decisions.
This study introduces goal-congruent outcomes as a novel mediator in ECM. Theoretically, this finding expands ECM by emphasizing that goal-congruent outcomes can foster stronger long-term user retention, even when users are no longer actively evaluating the system's utility. From a managerial perspective, digital platform designers and service providers should focus not only on enhancing system functionality but also on aligning system features with users' personal goals.
