Numerous species of Earth's biota are at risk of extinction and wildlife conservation is more important than ever. Reliable baseline data are essential for wildlife management to inform on the numbers and distribution of endangered species. A promising non-invasive and cost-effective method for monitoring endangered species is the Footprint Identification Technology (FIT). It lends itself to both conservation research as well as citizen science and can be combined with other data collection methods. FIT extracts and uses morphometrics from animal footprints to create geometric profiles that are analyzed in customized classification models. These can identify species, sex and individual. The ability to identify individuals can then be used to predict various population parameters including size, distribution, and growth rate. FIT has been developed and published for several species, but it requires high quality footprints. Perfect prints are not always easy to find in the field as various factors can influence their quality. In this paper, we demonstrate that geometrical profiles derived from poor quality footprints can be seen as datasets with missing values. Missing values are a common problem in various disciplines, and well-established strategies to impute missing values are widely available. We conducted two experiments to see whether such an approach could widen the application of the FIT method. The experiments were designed to test the hypothesis that population sizes can be underestimated when incomplete footprints are discarded from the data. We artificially introduced different proportions of missing values in datasets with geometric profiles of five different species for which FIT models have been published. We also analyzed a new dataset of geometric profiles of cheetah (Acinonyx jubatus) footprints not meeting the standard FIT requirements. We demonstrated that excluding incomplete footprints led to an underestimation of the known population. As an alternative to discarding footprints, we compared different imputation techniques as data pre-processing steps by comparing the performance of resulting FIT models.When imputation was chosen instead, we could show that FIT models with imputed geometric profiles were not significantly less accurate in predicting individual ID or population size even with high rates of missing values. We believe that our findings can be generalized, and the results indicate that imperfect footprints can contribute to the robustness of the FIT method and that this approach is particularly applicable when few good-quality footprints are available. We therefore highly recommend including imputation of imperfect footprints as a data pre-processing step.