Abstract
This study investigates how reputational intelligence derived from social media influences the financial performance of South Africa’s leading retail banks, focusing on First National Bank (FNB) and Capitec between 2020 and 2025, and provides a framework for integrating digital sentiment insights into strategic decision-making. In an era of increasing digital engagement, customer perceptions expressed online serve as critical indicators of brand reputation and business effectiveness. In order to analyse social media sentiment, the study uses five machine learning models: Passive Aggressive Classifier (PAC), Support Vector Machine (Linear SVM), Random Forest (RF), and Multinomial Naïve Bayes (MNB) and Logistic Regression (LR). Accuracy, recall, precision, F1-score, and Cohen's Kappa are used to evaluate the model's performance. With an accuracy of 89%, recall of 87.2%, and F1-score of 87.3%, PAC fared better than the other models, successfully capturing both explicit and subtle feelings, including neutral and ambiguous messages. The findings illustrate the strategic importance of reputational information in building customer trust, engagement, and loyalty by showing a substantial association between favourable sentiment trends and financial outcomes. FNB’s digital optimisation and Capitec’s combined physical-digital expansion exemplify how different strategic approaches can leverage real-time sentiment insights to drive growth. The study contributes theoretically by linking sentiment analysis with reputation and stakeholder theory and offers a practical framework for embedding sentiment analytics into business intelligence systems, supporting proactive risk management, customer-focused service alignment, and data-driven strategic planning. Overall, the findings underscore that social media-derived sentiment is a vital tool for transforming intangible reputational assets into measurable business value, enabling agile, informed, and customer-centric decision-making in a competitive financial landscape.