DEVELOPMENT OF AN AI-BASED SYSTEM FOR GENERATING MARKETING CREATIVES FOR CAMPAIGNS AND PROMOTIONS

ABSTRACT

The aim of this study is to develop an artificial intelligence-based system for automatically generating marketing creatives, including text and visual elements for advertising campaigns. The methodology involves the use of large language models for slogan generation and diffusion models for image creation, within a modular architecture that supports dynamic prompt construction and layout composition for various platforms. As a result, the system reduced creative production time from several hours to 20 minutes and delivered high-quality content: 85% of texts and 78% of visuals were deemed ready for publication. The findings confirm that integrating AI into marketing workflows enhances efficiency and creative flexibility, while also emphasizing the importance of human oversight and ethical responsibility when using generative models.

АННОТАЦИЯ

Цель исследования - разработка системы на основе искусственного интеллекта для автоматической генерации маркетинговых креативов, включая тексты и визуальные элементы рекламных кампаний. Методология включает использование больших языковых моделей для генерации слоганов и диффузионных моделей для создания изображений, а также модульную архитектуру с динамической настройкой подсказок и компоновкой макетов под разные платформы. В результате система сократила время создания креативов с нескольких часов до 20 минут и обеспечила высокое качество контента: 85% текстов и 78% изображений признаны готовыми к публикации. Выводы подтверждают, что внедрение ИИ в маркетинговые процессы повышает эффективность и креативную гибкость, при этом подчеркивается важность человеческого контроля и этической ответственности при использовании генеративных моделей.

Ключевые слова: генеративный ИИ, маркетинговые креативы, автоматическая генерация контента, ИИ в рекламе, построение подсказок, синтез изображений, персонализация кампаний, автоматизация креативных процессов.

Keywords: generative AI, marketing creatives, automated content generation, AI in advertising, prompt engineering, image synthesis, campaign personalization, creative workflow automation.

Introduction

The rise of artificial intelligence is fundamentally reshaping how businesses approach marketing. Traditional methods of producing promotional content are often time-consuming, resource-intensive, and limited in scalability. As marketing becomes increasingly data-driven and customer-centric, there is a growing demand for tools that can automate and personalize content creation without compromising quality or creativity.

One promising area of innovation is the use of generative AI to produce marketing creatives, including slogans, images, and campaign materials. These AI systems offer the potential to generate tailored content based on brand voice, target audience, and promotional goals freeing marketers from routine tasks and allowing them to focus on strategic and creative oversight.

Despite the promise, integrating AI into creative workflows is not without challenges. Issues such as content authenticity, ethical use of data, and the risk of generating biased or irrelevant content must be carefully managed. Moreover, marketing professionals must adapt to new tools and processes, balancing automation with human creativity.

This study presents the development of an AI-based system designed to generate marketing creatives automatically. The proposed system combines large language models and image generation algorithms to produce campaign-ready content. The paper outlines the architecture, methodology, testing framework, and future development paths, with the aim of demonstrating how AI can enhance marketing efficiency, reduce production costs, and enable greater creative agility.

Background and related works

AI in Marketing Practice

The application of AI in marketing has evolved significantly, transforming areas such as targeted advertising, dynamic content creation, and audience analysis (Park, Lee, & Kim, 2024). Modern AI tools leverage natural language processing and machine learning to enable large-scale content production and personalized messaging (Babadoğan, 2024). These technologies automate routine tasks including keyword optimization and performance tracking, allowing marketing teams to focus on strategic initiatives (Babadoğan, 2024).

Generative AI for Creative Content

Generative AI models like GPT-3 and DALL-E represent a fundamental shift in content creation capabilities. Hartmann et al. (2024) demonstrate how these systems can produce original marketing assets while maintaining brand consistency. However, outputs require careful validation for relevance and authenticity, particularly in sensitive applications.

Theoretical Underpinnings

Adstock theory provides a framework for understanding advertising's cumulative effects (Gijsenberg, Van Heerde, & Verhoef, 2011). This theoretical perspective informs AI system design, particularly in content sequencing and campaign timing. Kshetri, Amankwah-Amoah, and Ifinedo (2023) further examine how generative AI extends traditional marketing models through automated content production.

Ethical Considerations

Key challenges include data privacy compliance (GDPR, CCPA) and algorithmic bias mitigation (Demsar, Davison, & Hsu, 2025). Demsar et al. (2025) emphasize that transparency about AI's creative role remains essential for consumer trust, particularly when generating persuasive content.

Methodology

Problem Statement and Objectives

Traditional marketing creative production is slow, costly, and lacks scalability for personalized content. This research develops an AI system to:

• Automate generation of slogans, visuals, and full ads.
• Reduce production time/costs while maintaining brand consistency.
• Enable rapid customization for platforms/audiences.

System Architecture Overview

The system follows a modular design for flexibility and scalability. A user-friendly interface collects campaign inputs (product details, target audience, tone), which are processed by a dynamic prompt builder to generate AI instructions. Two core AI engines handle text (slogans, CTAs) and image generation, while a layout assembler structures these elements into cohesive ads. Finally, an export module delivers platform-optimized creatives (e.g., social media banners, web ads) in standard formats. This architecture supports batch processing, A/B testing, and seamless updates to individual components. Key processes:

• Prompt Engineering: Iterative refinements based on feedback/metrics to balance creativity and commercial viability.

• Layout Composition: Templates and responsive algorithms ensure readability, visual balance, and brand compliance.

Evaluation Metrics and Testing Setup

Output quality is assessed through automated checks (grammar, resolution) and human evaluations by marketing professionals, who rate originality, brand fit, and effectiveness. Efficiency gains are quantified by comparing AI-generated output to manual workflows. Failed cases are analyzed to identify systemic weaknesses, driving iterative improvements in prompts, models, and layout logic. Table 1 outlines how manual and AI-generated creatives perform across several quality dimensions.

Table 1.

Output Quality Evaluation Matrix

Criteria such as brief accuracy, visual appeal, originality, speed, and adaptability are rated on a scale, with accompanying comments for context. While manual methods slightly outperform in creativity and context alignment, the AI approach excels in speed and format versatility.

Results

Creative Quality Evaluation

The AI-generated slogans, taglines, and calls-to-action were tested for grammar and readability (using tools like Flesch Reading Ease). Over 85% met grammatical and advertising standards. A panel assessed 60 creatives across five industries (tech, retail, travel, healthcare, food & beverage), rating them on clarity, emotional appeal, call-to-action strength, and brand alignment. Average scores were 4.1 (clarity) and 3.9 (brand alignment), with slight tone inconsistencies in niche sectors. 78% of generated images were publication-ready, while 22% had minor flaws like awkward composition or unclear focus—often due to vague prompts.

Figure 1 visually compares the time spent on different stages of marketing creative development when performed manually versus using an AI-based system.

Figure 1. Workflow Comparison: Manual vs. AI-Powered Creative Generation

Manual workflows consume significantly more time, especially during concept development and design phases, which are reduced to near-instant tasks in the AI-driven process.

Diversity and Originality Assessment

• Text Outputs:

  • o N-gram analysis: <20% overlap between variations.

  • o Human evaluation: 84% of slogans rated "fresh and campaign-ready."

• Visual Outputs:

  • o Varied color palettes, compositions, and contextual elements.

  • o Strong diversity in lifestyle categories (e.g., product interactions, environmental scenes).

Efficiency Gains and Time Reduction

• Time Reduction:
  • o 90% faster: 20 mins (AI) vs. 3–6 hours (manual).
  • o Enables 80% reallocation of creative time to strategy/engagement.
• Iteration Speed:
  • o Rapid A/B testing (multiple variations in minutes).
  • o Eliminates bottlenecks, shifting focus to high-value tasks.

Conclusion

This research demonstrates how AI can transform marketing creative development by automating content generation while maintaining quality. The developed system successfully produces campaign-ready text and visuals that align with brand standards, reducing production time from hours to minutes. Evaluation results confirm its ability to generate diverse, engaging content across product categories - with text outputs scoring well for clarity and visuals meeting professional standards in most cases.

However, the system works best as a collaborative tool rather than a full replacement for human creativity. Certain outputs, particularly for niche products or abstract concepts, still require human refinement. Prompt quality remains crucial, emphasizing the need for ongoing optimization. Ethical considerations around bias and transparency also demand attention as these systems evolve.

Looking ahead, expanding into dynamic formats (animation, interactive content) and multilingual support could further enhance the system’s value. Future development should prioritize real-time adaptation features, automated quality checks, and bias mitigation tools. When thoughtfully integrated, AI-powered creative systems enable marketers to focus on strategy and curation - combining algorithmic efficiency with human insight to drive more personalized, agile campaigns.

References:

1. Aurier, P., & Broz-Giroux, A. (2013). Modeling advertising impact at campaign level: Empirical generalizations relative to long-term advertising profit contribution and its antecedents. Marketing Letters, 25(2), 193. https://doi.org/10.1007/s11002-013-9252-3
2. Babadoğan, B. (2024). Exploring the Role of AI in Automating Content Marketing: Unlocking Opportunities and Navigating Challenges. Next Frontier., 8(1), 67. https://doi.org/10.62802/gkj6f352
3. Demsar, V., Ferraro, C., Sands, S., & Kohn, A. (2025). Harmony or Discord? The Intersection of Generative AI and Human Creativity in Advertising. Journal of Advertising Research, 1. https://doi.org/10.1080/00218499.2025.2464305
4. Gijsenberg, M. J., Heerde, H. J. van, Dekimpe, M. G., & Nijs, V. R. (2011). Understanding the Role of Adstock in Advertising Decisions. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.1905426
5. Hartmann, J., Exner, Y., & Domdey, S. (2024). The power of generative marketing: Can generative AI create superhuman visual marketing content? International Journal of Research in Marketing. https://doi.org/10.1016/j.ijresmar.2024.09.002
6. Kshetri, N., Dwivedi, Y. K., Davenport, T. H., & Panteli, N. (2023). Generative artificial intelligence in marketing: Applications, opportunities, challenges, and research agenda. International Journal of Information Management, 75, 102716. https://doi.org/10.1016/j.ijinfomgt.2023.102716
7. Park, H. W., Lim, C. V., Zhu, Y. P., & Omar, M. (2024). Decoding the Relationship of Artificial Intelligence, Advertising, and Generative Models. https://doi.org/10.20944/preprints202401.0373.v1