"background": "Manufacturing systems in Kenya face persistent efficiency challenges, yet rigorous field-based diagnostic methodologies tailored to the local industrial context are scarce. Existing approaches often rely on retrospective data or theoretical models, lacking causal rigour for identifying actionable improvements. ", "purpose and objectives": "This article presents a novel methodological framework for conducting randomised field trials (RFTs) to diagnose and quantify efficiency gains within manufacturing plants. The objective is to provide a structured, evidence-based protocol for engineers to implement controlled interventions and measure their causal impact on system performance. ", "methodology": "The proposed RFT methodology involves a multi-stage protocol: (1) system characterisation and bottleneck identification, (2) random assignment of participating plants to treatment or control groups, (3) implementation of targeted technical interventions in treatment groups, and (4) longitudinal data collection on key performance indicators. The core analysis employs a difference-in-differences model: Y{it = \0 + \1 + \2 + \ (\) + \₈ₓ, where inference is based on cluster-robust standard errors at the plant level. ", "findings": "As a methodology article, this paper presents no empirical results from a specific trial. However, the framework is designed to detect effect sizes with a minimum detectable effect of 15% improvement in overall equipment effectiveness (OEE) with 80% power, given typical variability observed in pilot data. The diagnostic process prioritises energy and material flow metrics as primary indicators. ", "conclusion": "The structured RFT methodology provides a robust, causal framework for efficiency diagnostics, moving beyond correlational analysis. It is specifically designed to address common data quality and implementation challenges in the local manufacturing environment. ", "recommendations": "Practitioners should adopt this RFT protocol for pilot studies before full-scale plant investments. Researchers are encouraged to utilise the framework to
Otieno et al. (Tue,) studied this question.