Estimating trends in population size is critical for understanding population status and assessing the success of management interventions. Visual counts of birds as they congregate around predictable locations, such as waterholes, is a popular technique for estimating population size. Bird counts are used as a proxy for abundance, but how the relationship between counts and actual abundance varies over space and time is rarely assessed. Here, we demonstrate that colour banding and motion detection cameras provided a good method for monitoring finch visitation patterns across space and time. These methods were validated using three sympatric finch species, the abundance of which have been estimated from waterhole counts over many years. The study showed significant temporal inter-species variability in the proportion of birds visiting waterholes and the number of times the same individual returned to the same waterhole during the early morning. Bird visitation rates also varied between consecutive days, across adjacent waterholes and at different stages of the dry season. Our study suggests that spatiotemporal variation in individual behaviour may introduce substantial error into site-focused bird counts and we recommend considering this in census design.