Cost-effective methods, such as dung counts, are widely used for monitoring wildlife population abundances but often yield estimates with low precision and wide confidence intervals. In this study, we assess the impact of different statistical analyses---traditional mathematical approaches, bootstrapping, and Bayesian modelling---on the precision and accuracy of population estimates for red and roe deer on Scotland's west coast. Both bootstrapping and Bayesian modelling reduced estimate uncertainty compared to traditional methods, providing more precise estimates. Bayesian modelling further accounted for the overdispersion characteristic of dung count data, offering a more ecologically robust and statistically sound approach to estimating population densities.