Integrating is success and open innovation in enterprise LLM adoption: a dual-path model of net benefits and usage growth

Abstract

Firms are increasingly adopting Large Language Models (LLMs) to enhance productivity and eventually save costs and increase revenue. Continued LLM adoption requires that LLMs demonstrate tangible Return-on-Investment, making it essential to verify whether cost reductions and revenue contribution stem from the quality of the LLMs or from organisational capabilities in learning from LLMs. Based on the theories of information success, open innovation, and organisational learning, this paper examines how firms’ evaluations of LLM quality and their learning and innovation capabilities shape financial performance perceptions, which in turn drive increased utilisation of LLMs. A post-hoc experiment was conducted to reexamine the relationship between information quality and user satisfaction with LLMs. The survey sample consisted of 261 firm representatives (one key informant per firm). For the post-hoc experiment, 148 corporate employees with prior experience utilising LLM-based services were recruited. Evaluation of the measurement properties, structural model testing, and post-hoc analyses were performed using SPSS-29.0, PLS-SEM-4.1, and Lavaan package (v0.6-15). The results reveal a finding that challenges conventional assumptions about how information quality shapes user satisfaction in LLM contexts. This study aims to provide practical implications for firms seeking to strengthen the organisational foundations necessary for sustainable value creation through LLM adoption.