Scholarlink

<p>Customer churn poses a growing risk for public sector banks in emerging economies as customers shift toward private and new-generation banks that offer stronger digital services. This study examines the antecedents of churn in the State Bank of India through a dual-method design that integrates TabNet deep learning with fuzzy-set Qualitative Comparative Analysis (fsQCA). The dataset consists of survey responses from 611 SBI customers and covers branch, mobile banking, and internet banking quality, trust, and reported service problems. Reliability and validity checks confirm that the measurement structure supports further analysis. The TabNet model records strong predictive performance, and its attention weights and SHAP outputs show that internet banking quality stands out as the strongest contributor to churn risk, followed by mobile banking quality and service problems. Branch quality and trust make smaller contributions. The fsQCA results identify several pathways that lead to high churn. High service problems together with low internet banking quality form the most consistent pathway. Moderate problem levels combined with weak trust and low branch quality also increase churn. Serious problems with poor mobile banking quality raise churn even when trust remains higher. These findings indicate that improvements in digital service quality and clear problem-resolution systems should form the core of the retention strategy. Trust and satisfaction do not offset the effects of operational failures or weak digital performance.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Key drivers of customer churn in public sector banks in India: evidence from the State Bank of India customers

  • 期刊论文

  • 发表日期：

    2026年3月12日

查看全文
  • Mohanan Moni,
  • Ganesh B. Nair,
  • Sreeraj Venuraj,
  • Greeshma Sajan

摘要

Customer churn poses a growing risk for public sector banks in emerging economies as customers shift toward private and new-generation banks that offer stronger digital services. This study examines the antecedents of churn in the State Bank of India through a dual-method design that integrates TabNet deep learning with fuzzy-set Qualitative Comparative Analysis (fsQCA). The dataset consists of survey responses from 611 SBI customers and covers branch, mobile banking, and internet banking quality, trust, and reported service problems. Reliability and validity checks confirm that the measurement structure supports further analysis. The TabNet model records strong predictive performance, and its attention weights and SHAP outputs show that internet banking quality stands out as the strongest contributor to churn risk, followed by mobile banking quality and service problems. Branch quality and trust make smaller contributions. The fsQCA results identify several pathways that lead to high churn. High service problems together with low internet banking quality form the most consistent pathway. Moderate problem levels combined with weak trust and low branch quality also increase churn. Serious problems with poor mobile banking quality raise churn even when trust remains higher. These findings indicate that improvements in digital service quality and clear problem-resolution systems should form the core of the retention strategy. Trust and satisfaction do not offset the effects of operational failures or weak digital performance.