In contrast to the goals of the symposium from which this series of papers originated, we argue that attempts to apply unambiguously defined and general management unit criteria based solely on genetic parameters can easily lead to incorrect management decisions. We maintain that conservation genetics is best served by altering the perspective of data analysis so that decision making is optimally facilitated. To do so requires accounting for policy objectives early in the design and execution of the science. This contrasts with typical hypothesis testing approaches to analysing genetic data for determining population structure, which often aspire to objectivity by considering management objectives only after the analysis is complete. The null hypothesis is generally taken as panmixia with a strong predilection towards avoiding false acceptance of the alternative hypothesis (the existence of population structure). We show by example how defining management units using genetic data and standard scientific analyses that do not consider either the specific management objectives or the anthropogenic risks facing the populations being studied can easily result in a management failure by losing local populations. We then use the same example to show how an ‘applied’ approach driven by specific objectives and knowledge of abundance and mortality results in appropriate analyses and better decisions. Because management objectives stem from public policy, which differs among countries and among species groups, criteria for defining management units must be specific, not general. Therefore, we conclude that the most productive way to define management units is on a case-by-case basis. We also suggest that creating analytical tools designed specifically to address decision making in a management context, rather than re-tooling academic tools designed for other purposes, will increase and improve the use of genetics in conservation.