CRITIC-DEMATEL Framework: Optimizing and Reshaping Digital Marketing Content Strategy with Large Language Models

A latest study proposes the CRITIC-DEMATEL two-stage framework to help marketers systematically optimize marketing content generated by large language models (LLMs). Combining two multi-criteria decision-making methods—CRITIC and DEMATEL—this framework identifies 15 key influencing factors, offering new insights for SEO strategies and digital marketing content creation in the AI era.

With the popularity of LLMs like ChatGPT and Claude, marketers have gained efficient productivity tools, but simply using AI to generate content cannot guarantee results. How to balance content quality, SEO requirements, and target audience reach has become a core issue in the industry. Traditional SEO relies on keyword stuffing and external link building, which can no longer meet users' demand for high-quality, in-depth content in the LLM era.

The research team combined CRITIC and DEMATEL to build a two-stage framework: CRITIC is used to evaluate the importance weights of various influencing factors, while DEMATEL reveals the causal relationships and dependencies between factors. The two-stage design forms an organic whole, allowing decision-makers to not only know 'what is important' but also understand 'why it is important' and 'how factors interact with each other'.

Through expert interviews and literature analysis, the study identified 15 core factors affecting the effectiveness of LLM Optimization (LLMO), covering three dimensions: technology, content, and user experience:

• Technical dimension: Quality of Retrieval-Augmented Generation (RAG), degree of model fine-tuning, level of prompt engineering, etc.
• Content dimension: Originality, readability, depth of information, clarity of structure, etc.
• User experience dimension: Relevance, practicality, credibility, visual presentation, etc. There are complex interactions between factors—for example, technical optimization indirectly affects content readability, and originality influences search engine weight distribution.

The causal network graph output by DEMATEL can identify 'cause-type' factors (which widely affect other factors) and 'result-type' factors (which are greatly influenced by other factors). The study found that investment in technical infrastructure and team AI literacy are cause-type factors—prioritizing these investments can produce chain reactions; short-term traffic growth is a result-type factor, and excessive focus on it can lead to strategic short-sightedness. This graph helps marketing teams shift from passive response to systematic forward-looking planning.

The framework provides guidance for different roles:

• Content creators: Shift from 'intuition-based creation' to 'data-driven creation', and regularly evaluate the 15 factors to identify optimization entry points.
• Marketing agencies: Use analysis results to demonstrate the scientific nature of strategies, enhance client trust, and achieve service standardization and effect evaluation.
• Enterprise decision-makers: Need to establish a systematic content optimization system and cultivate the team's AI application capabilities, rather than simply purchasing AI tools.

LLMO is moving from the edge to the mainstream, and this study provides the first systematic decision support framework. More interdisciplinary research (integration of operations research, AI, and marketing) will emerge in the future. Practitioners need to pay attention to cutting-edge research and transform it into practical tools; the core remains to provide real value to users—technology is an amplifier, and the key to success lies in the depth of understanding of user needs and the ability to translate that into high-quality content.