This case study presents a rigorous, end-to-end data preparation workflow for Likert-scale survey research, highlighting the critical steps required before any statistical analysis can be considered valid. Using a simulated dataset of 353 public sector employees in Gauteng, the study examines Perceived Supervisor Support, Burnout, and Intention to Leave. It demonstrates best practices in data cleaning, descriptive analysis, normality testing, and reliability and validity assessment, ultimately producing a clean, defensible dataset with strong internal consistency across all constructs. Designed as a practical template, it equips postgraduate students with a clear, structured approach to preparing survey data that can withstand academic scrutiny.

A South African SME's FY 2024 revenue report overstated income by R2,985,788 — caused by four entirely preventable data quality issues. Seven structured cleaning steps reduced the error rate from 72.1% to zero, giving the business its first reliable view of actual performance.