Techniques from the world of marketing could be applied by conservationists to improve behaviour change outcomes. One example is "audience segmentation", a method to improve targeting of interventions by differentiating groups that exist within a population. Tropical forest conservation typically involves a combination of different tools, requiring managers to make site-specific decisions about which to use and where. To design the most effective interventions, there first needs to be a clear understanding of who the intervention intends to influence. However, this is often not well defined in conservation projects, which are typically aimed at broad groups such as "local communities". This one-size-fits-all approach is inefficient if the population is comprised of heterogeneous groups that respond differently to different interventions. An improvement could be to apply the marketing technique of audience segmentation. Segmentation methods aim to sub-divide populations into groups that are internally homogenous, but differ from each other, in their response to different behaviour change mechanisms. This provides a basis for designing targeted interventions that are optimal for a specific group or groups. We evaluate audience segmentation as a tool in tropical forest conservation, using a case study from the Gola Forest, Liberia. We ask the pragmatic question of whether segmentation would have practical advantages over simpler approaches, given that conservation practitioners rarely have access to large datasets and little is known about the factors linked to successful behaviour change. We apply infinite binomial mixture models to perform cluster analyses of simple datasets describing household and individual livelihoods respectively, then compare the management implications to simpler approaches of using hunting behaviour as the targeting criterion. We found that targeting based only on hunting behaviour did not greatly improve efficiency of decisions relative to one-size-fits-all. The segmentation approach identified a distinct set of priorities for livelihood support tools relative to simpler methods, and provided novel insight for managers. A more nuanced perspective on targeted intervention design was possible by characterising segments across multiple traits. However, this advantage was constrained by the traits used to define segments which represented basic livelihood patterns, rather than psychographic factors associated with behaviour change. Segments captured variation in 3 out of 4 traits taken to be indicators of behavioural response to interventions such as law enforcement and messaging campaigns, whereas simple target group definitions did not. This provides evidence that segments would be a valid basis for designing targeted interventions. We conclude that even under conditions of limited data availability, audience segmentation is a promising tool to guide intervention design for site-based programmes.