Hypothesis: \nCoating-free approaches to achieve liquid repellent, or omniphobic, surfaces could exploit inexpensive intrinsically wetting materials, such as polyethylene terephthalate and nylon, for applications such as liquid-vapor extraction and drag reduction. However, it is not clear whether the existing criteria for assessing coating-based omniphobicity, based on apparent contact angles, would be reliable for coating-free approaches, especially considering localized defects/damages during manufacturing and usage. \nExperiments: \nWe assessed the omniphobicity of silica surfaces adorned with arrays of doubly reentrant pillars, cavities, and hybrid designs with sessile drops and on immersion in water and hexadecane through contact angle goniometry and confocal microscopy, respectively. \nFindings: \nWe demonstrate that the assessment of omniphobicity of surfaces derived from intrinsically wetting materials can be misleading, if solely based on the measurement of contact angles. Specifically, localized defects in microtextures consisting of pillars may lead to the spontaneous loss of omniphobicity and detecting them through contact angles can be difficult. We also demonstrate that the immersion of those surfaces into probe liquids may serve as a simple and quick ‘litmus’ test for omniphobicity. Thus, immersion as additional criterion for omniphobicity might prove itself useful in the context of large-scale manufacturing.