Customer Service Channels and Response Time in Online Retail: Global Insights with Evidence from Bangladesh

Abstract

This thesis examines how customer service channel choice and response time shape customer satisfaction in Bangladesh’s online retail, comparing local and international providers and interpret-ing results through SERVQUAL, Expectation–Confirmation Theory and the Technology Acceptance Model. Two structured, self-administered Google Forms surveys were deployed—customers (n = 70) and retail staff (n = 30)—and the realized data were analyzed using descriptive statistics, cross-tabulations and targeted tests where cell counts permitted, implemented mainly in Excel with selective Python for reproducibility. Messaging channels, especially WhatsApp, concentrated minutes and sub-hour first responses and were associated with higher resolution and more positive expectation confirmation. Customers most often perceived phone and live chat as fastest, while staff rated social media as the most efficient to handle, signaling an operation–perception gap. Formal SLAs were not universal (43%), yet many teams still reported sub-hour first replies; first-contact resolution was strong for a major-ity (≥76% in 63%). Delays carried behavioral costs, with 46% of customers reporting at least one purchase abandoned due to slow support. On perceived performance, 47% judged local providers worse than international peers, though localization levers—Bengali language support, cash-on-delivery, domestic returns and local-hour coverage—mitigated risk, accelerated replies and raised assurance when executed well. This thesis contributes Bangladesh-specific evidence on channel–timing mechanisms, integrates SERVQUAL, ECT and TAM into a parsimonious pathway from response speed and channel fit to satisfaction and continuance, and offers a dual-stakeholder view that triangulates customer expe-rience with staff operations. Practically, the thesis recommends publishing channel-specific SLAs, prioritizing WhatsApp and live chat coverage in peak windows, adopting lightweight ticketing and live-chat tools, training agents to signal ownership and timelines, and tracking first response, FCR and abandonment by channel. Limitations include non-probability sampling, closed-ended measures, self-reported timings and a single-country cross-section. Future research should pair probability sampling with system telemetry, and A/B test SLA thresholds and staffing models to quantify causal impacts on satisfaction and loyalty.