Cosmic voids defined as a subset of Delaunay triangulation (DT) circumspheres have been used to measure the baryon acoustic oscillations (BAO) scale; providing tighter constraints on cosmological parameters when combined with matter tracers. These voids are defined as spheres larger than a given radius threshold, which is constant over the survey volume. However, the response of these void tracers to observational systematics has not yet been studied. In this work, we analyse the response of void clustering to selection effects. We find for the case of moderate (<20 per cent) incompleteness, void selection based on a constant radius cut yields robust measurements. This is particularly true for BAO-reconstructed galaxy samples, where large-scale void exclusion effects are mitigated. Moreover, we observe for the case of severe (up to 90 per cent) incompleteness - such as can be found at the edges of the radial selection function - that an accurate estimation of the void distribution is necessary for unbiased clustering measurements. In addition, we find that without reconstruction, using a constant threshold under these conditions produces a stronger void exclusion effect that can affect the clustering on large scales. A new void selection criterion dependent on the (local) observed tracer density that maximizes the BAO peak significance prevents the aforementioned exclusion features from contaminating the BAO signal. Finally, we verify, with large simulations including light-cone evolution, that both void sample definitions (local and constant) yield unbiased and consistent BAO scale measurements.