Work Engagement and Turnover Intention among Customer-Support Executives at a Hyderabad BPO (UWES-9, SPSS)

Attrition is the single most expensive HR problem in Indian business-process outsourcing: annual attrition of 30–45% in voice processes is routinely reported, and every exit means recruitment, three to four weeks of process training and weeks of lower productivity. HR teams usually respond with pay hikes and exit interviews, but exit interviews come too late and tell you why people left, not who is about to leave.

This internship project measures work engagement and turnover intention among customer-support executives at Deccan CX Services, a fictional 620-employee BPO in Hyderabad running inbound voice, chat/email and back-office processes. Using a structured questionnaire built on two established instruments — the Utrecht Work Engagement Scale (UWES-9) and the six-item Turnover Intention Scale (TIS-6) — plus short scales for perceived supervisor support and pay satisfaction, the study surveys a stratified random sample of 243 employees.

In SPSS, the project tests whether engagement differs by shift and tenure (t-test, ANOVA), whether engagement and turnover intention are correlated, whether the process worked in is associated with high quit intention (chi-square), and which factors best predict turnover intention (multiple regression). The output is a set of evidence-based retention actions for the HR manager and a report aligned with the AICTE model MB 303 Internship Project & Viva (report 50, presentation 30, viva 20).

Abstract

High attrition in Indian BPOs raises hiring and training costs and harms service quality. This study examines the relationship between work engagement, perceived supervisor support, pay satisfaction and turnover intention among customer-support executives of a Hyderabad BPO. A stratified random sample of 243 employees completed a questionnaire containing UWES-9, TIS-6 and supporting scales. Reliability was checked with Cronbach's alpha, and hypotheses were tested using independent-samples t-test, one-way ANOVA, Pearson correlation, chi-square and multiple regression in SPSS. The findings are converted into targeted retention actions.

Introduction

Engagement, in the Schaufeli–Bakker sense, is a positive work-related state of vigour, dedication and absorption. Engaged employees are less likely to leave, but engagement itself depends on the job's demands and resources — shift timings, call targets, supervisor behaviour and pay. In BPOs these conditions vary sharply between voice and non-voice processes and between day and night shifts, which makes the setting well suited to comparison.

Review of literature (summary)

International studies consistently find a negative relationship between engagement and turnover intention, and a positive relationship between supervisor support and engagement (job demands–resources model). Indian studies on BPO attrition focus heavily on pay, night shifts and career growth, often using self-made questionnaires with unreported reliability. Gap: few Indian BPO studies use validated engagement and turnover-intention scales together with process- and shift-level comparisons in the same sample.

Existing vs proposed approach at the company

• Existing: monthly attrition report, exit interviews, annual pay benchmarking; no measure of engagement among current staff.
• Proposed: a validated, repeatable engagement and intention survey, statistical identification of high-risk segments, and a short list of actions with measurable targets.

Feasibility

• Operational: HR agreed to circulate the survey during team huddles; each response takes about 8 minutes.
• Technical: SPSS is available in the institute lab; Excel is used for coding and charts.
• Ethical: participation is voluntary and anonymous; no employee IDs are collected; HR receives only aggregated results; written permission from the HR head and informed consent from each respondent are recorded.

Deccan CX Services, a fictional BPO in Hyderabad with about 620 customer-support executives, recorded annualised attrition of roughly 38% last year, concentrated in inbound voice processes and night shifts. HR relies on exit interviews, which capture reasons only after resignation and are often diplomatic rather than honest. Management has approved a retention budget but does not know whether to spend it on pay, supervisor training, shift rotation or recognition.

The problem is the absence of reliable data on current employees' engagement and intention to quit, and on which workplace factors are most strongly associated with that intention. Without it, retention spending is based on guesswork. The study therefore measures work engagement and turnover intention with validated scales, compares them across shifts, processes and tenure groups, and identifies the strongest predictors of turnover intention, so that HR can target the segments and levers that matter most.

In scope

• Customer-support executives (non-managerial) at one Hyderabad BPO with at least one month's tenure, across three processes and two shift types.
• Variables: work engagement and its three dimensions, turnover intention, perceived supervisor support, pay satisfaction and demographics (age band, gender, education, tenure, shift, process).
• Cross-sectional survey during an eight-week internship; analysis in SPSS; recommendations to HR.

Out of scope

• Team leaders and managers; employees on notice period.
• Actual turnover follow-up (the study measures intention, not later exits).
• Salary data from payroll — pay is measured only as perceived satisfaction.
• Other BPOs; results are specific to this organisation and generalise only cautiously.

Research design

Descriptive and causal-comparative (correlational) cross-sectional survey.

Population, sampling and sample size

Population N = 620 executives (voice 340, chat/email 170, back office 110). Using Yamane's formula n = N / (1 + N·e²) with e = 0.05: n = 620 / (1 + 620 × 0.0025) = 620 / 2.55 ≈ 243. Proportionate stratified random sampling by process: voice 133, chat/email 67, back office 43. Within each stratum, respondents were drawn at random from the HR roster using Excel's RAND() function; non-responders were replaced from the same stratum. Target response was 260 to allow for incomplete forms.

Instrument

A pilot with 30 employees (excluded from the main sample) checked wording and reliability. Cronbach's alpha ≥ 0.70 was required for every scale; a bilingual (English/Telugu) version was offered.

Hypotheses (α = 0.05)

• H1₀: no difference in mean engagement between day and night shifts — independent-samples t-test (Levene's test first).
• H2₀: no difference in mean engagement across tenure groups (<6 months, 6–24 months, >24 months) — one-way ANOVA, Tukey post-hoc.
• H3₀: no correlation between engagement and turnover intention — Pearson correlation.
• H4₀: process and turnover-intention category are independent — chi-square test of independence, Cramér's V for effect size.
• H5₀: engagement, supervisor support and pay satisfaction do not jointly predict turnover intention — multiple linear regression (enter method), with VIF, Durbin–Watson and residual checks.

Timeline (8 weeks)

Week 1 orientation and HR discussions · Week 2 literature review and instrument · Week 3 pilot and reliability · Weeks 4–5 data collection · Week 6 coding and SPSS analysis · Week 7 findings discussion with HR · Week 8 report and presentation.