The exponential growth of social media platforms has transformed the way customers interact with brands, making actual sentiment analysis (SA) an essential component in shaping effective marketing strategies. However, existing models often lack robustness in processing high-volume, emotion-rich content dynamically. To address such a gap, a novel real-time framework is suggested for customer SA, specifically designed for social media marketing optimization. The Dynamic Coot Bird algorithm-mutated Adaptive Random Forest Classifier (DCB-ARFC) is designed to improve sentiment classification accuracy and adaptability. Publicly available datasets containing labeled social media posts from diverse platforms are used, ensuring relevance to marketing contexts. Data preprocessing is performed using a text tokenization and padding normalization approach, followed by the removal of noise and redundant text by stop word removal. For feature extraction, Term Frequency–Inverse Document Frequency (TF-IDF) and word embeddings are employed to capture contextual sentiment cues. The processed features are passed through the DCB-ARFC model, where the DCB algorithm optimized hyperparameters dynamically, while the ARFC ensured model resilience against real-time data fluctuations. The system demonstrated superior results in sentiment classification, attaining an accuracy of 0.992, a precision of 0.853, a recall of 0.837, and an F1-score of 0.842, significantly outperforming traditional static models. The framework enables marketers to adjust content strategies on the fly based on sentiment trends, enhancing user engagement and campaign responsiveness. Overall, the research offers a cutting-edge, intelligent framework for real-time sentiment-driven decision-making in digital marketing environments, contributing to academic insight and practical applications in the evolving field of Artificial Intelligence (AI)-powered social media analytics.