The complexity and opacity of modern digital advertising platforms, especially Meta Ads, have become major concerns in the marketing community. These platforms independently control key elements like audience targeting, pricing, and ad relevancy, operating in markets dominated by network effects. Companies such as Meta and Google hold leading positions, attracting many advertisers who often rely on intuition instead of data-driven insights. Consequently, billions are spent on social media ads that frequently fail to yield effective outcomes. Central to this issue are proprietary algorithms that utilize vast datasets unavailable to external advertisers. These algorithms act as black boxes, concealing how decisions about targeting and pricing are made. This lack of transparency severely limits advertisers’ ability to make informed choices, creating widespread demand for increased openness in the industry. In response, there is a growing movement advocating for transparency, industry-wide metric standardization, and stronger regulatory frameworks. These efforts aim to establish a fairer, more efficient digital advertising ecosystem, enabling advertisers to better assess campaign effectiveness and allocate budgets wisely. Beyond regulation, technological innovations are being pursued to equip marketers with advanced tools for navigating the digital advertising landscape. One promising approach involves machine learning models that predict click-through rates (CTR) for new ads, allowing marketers to evaluate potential engagement before large investments. However, marketers face another challenge: the overwhelming volume of data amid intense competition makes it difficult to extract meaningful insights or actionable strategies. The flood of data often causes information overload, leaving marketers uncertain about interpretation and application. To address this gap, researchers have turned to the success of large language models (LLMs) like ChatGPT, known for bridging complex technical data and non-technical audiences.

Inspired by this, a novel system named SODA (System for Optimal Digital Advertising) has been proposed. SODA combines LLMs with explainable artificial intelligence (AI) to improve interpretability of complex datasets and facilitate smooth collaboration between human marketers and AI. Specifically designed for digital marketing, SODA leverages LLMs’ natural language capabilities alongside explainable features, including advanced text-image models. This integration creates a user-friendly interface that simplifies data interpretation, helping marketers gain clearer insights into advertising metrics, competitor trends, and consumer behavior. By enhancing human-AI collaboration, SODA marks a significant advance in optimizing marketing strategies. Its ability to clarify complex data patterns and transparently explain AI-driven predictions empowers marketers to make confident, data-informed decisions. Moreover, by reducing the opacity of current advertising algorithms, SODA enables more effective and efficient use of advertising budgets. In summary, digital advertising faces significant challenges from opaque algorithms and vast, complex data. Dominant platforms like Meta and Google control much of the market through network effects, underscoring the urgent need for greater transparency and standardized metrics. Simultaneously, advances in machine learning and explainable AI offer promising solutions to help marketers harness abundant data and optimize strategies. Systems like SODA highlight the potential of integrating cutting-edge AI with marketing expertise to drive the digital advertising industry toward greater clarity, efficiency, and success.