Abstract
Corporate reputation has become a measurable asset that influences both financial performance and stakeholder confidence. In South Africa’s digital banking environment, reputational risk has intensified as online customer opinions increasingly shape market trust and stability. However, few studies have combined sentiment analytics with financial indicators in this context.
This study develops a data-driven reputational intelligence framework that predicts early warning signals of reputational risk and proposes data-informed mitigation strategies. The objectives include identifying key reputation drivers, quantifying their financial impact, and developing a predictive model and dashboard to support proactive risk management.
A mixed-method longitudinal design was applied using online reviews, public discussions, and financial data from January 2022 to August 2025. Automated web scraping collected customer feedback from digital platforms. Sentiment was analysed using a fine-tuned FinBERT model, and topics were classified using a twelve theme taxonomy. Weekly sentiment and topic metrics were aligned with financial indicators using regression and predictive models including Logistic Regression, Random Forest and XGBoost.
The framework identified five key drivers of reputational risk: transaction delays, fraud and security issues, credit and lending complaints, and service experience concerns. Together, these explained 80% of financial volatility (R² = 0.80). The XGBoost model achieved 88% accuracy, confirming that digital sentiment predicts financial outcomes. The framework turns reputation management into a proactive process through an automated dashboard that detects and mitigates reputational risks in South African banking.