Toward Effective AIGC for Marketing: A Theory-Driven System Design and Empirical Evaluation

Abstract

This paper introduces a theory-driven AIGC (AI-generated content) system for marketing image generation, grounded in visual marketing theory and implemented through structured prompt engineering, AI-based image generation, and a LLM-guided evaluation and selection process. The system employs a multi-agent architecture—comprising prompting, generation, and evaluation agents—to ensure content diversity, product authenticity, and theoretical alignment. Empirical evaluations across Meta Ads and Prolific show that the system significantly outperforms baseline AIGC—which lack theoretical grounding—in marketing effectiveness, and performs competitively with professionally generated content (PGC)—exceeding it in ad engagement while trailing in perceived effectiveness. The system also supports scalable theory validation through automated, controlled image generation. This work offers a practical and theoretically grounded framework for enhancing the reliability, adaptability, and research utility of generative AI in both commercial and academic contexts.