The core of the educational system is students' academic performance, which demands sensitive measures. In this situation, teacher-made tests (TMTs) are more promising, but they can be susceptible to measurement error if not well designed. Hence, this study aimed to investigate the relationship between the properties of TMTs and students' academic performance in Ethiopian public universities, evidencing English language communicative skills courses. The research employs a postpositivist research paradigm, quantitative approach, cross-sectional design, purposive sampling size technique, and CTTs. It revealed that the relationship that exists between the properties of TMTs and students' academic performance is strong positive, ranging from r = 0.773 to 0.886. This implies that the variance shared between these two variables was r[superscript 2] = 0.59 to 0.75/59% to 75%. A one-way ANOVA depicts (F[subscript 1.202] at p = 0.30>0.05) there was no statistically significant mean difference among public universities in students' academic performance, suggesting their homogeneity. In addition to this, in all public universities, students' score mean was less than the minimum requirement [mean of 37.42 out of 80, which is < 40 (50%) at P = 0.326 >0.05], needing concern. Design-thinking utilizing correlation enabled the creation of a novel model, while the limitation of correlation statistics was maintained by computing r[superscript 2]. The model, 'The Psychometric and Performance Properties of TMTs in Measuring Students' Academic Performance', requires further enhancement through interventional studies.