The Effect of LLM Automation on Performance Marketing Scale: Evidence on CAC, ROMI, and Creative Cycle Velocity

Performance marketing operations in technology-intensive firms have increasingly incorporated large language model (LLM) automation as a mechanism for addressing scalability constraints. This paper examines the quantitative relationship between the degree of LLM automation and three performance indicators: customer acquisition cost (CAC), return on marketing investment (ROMI), and the duration of the creative production cycle. A systematic review of peer-reviewed literature was combined with a comparative case analysis of five digital-native organizations that implemented LLM-based automation stacks between 2022 and 2024. The findings indicate that full LLM automation is associated with a CAC reduction of approximately 39 % relative to unaided baselines, a ROMI increase of 68 %age points, and a compression of the creative cycle from an average of 17.5 days to approximately 2.5 days. Furthermore, A/B test velocity increased by a factor of approximately eight, and the number of active audience segments expanded from fewer than ten to more than ninety. These results suggest that LLM automation constitutes a structural shift in performance marketing capacity rather than an incremental efficiency gain. The findings are of interest to marketing technology practitioners, digital product managers, and researchers studying AI-driven organizational transformation.