AbstractAs the extent of oil palm (Elaeis guineensis) cultivation has expanded at the expense of tropical rainforests, enriching conventional large‐scale oil palm plantations with native trees has been proposed as a strategy for restoring biodiversity and ecosystem function. However, how tree enrichment affects insect‐mediated ecosystem functions is unknown. We investigated impacts on insect herbivory and pollination in the fourth year of a plantation‐scale, long‐term oil palm biodiversity enrichment experiment in Jambi, Sumatra, Indonesia. Within 48 plots systematically varying in size (25–1600 m²) and planted tree species richness (one to six species), we collected response data on vegetation structure, understory insect abundances, and pollinator and herbivore activity on chili plants (Capsicum annuum), which served as indicators of insect‐mediated ecosystem functions. We examined the independent effects of plot size, tree species richness, and tree identity on these response variables, using the linear model for random partitions design. The experimental treatments were most associated with vegetation structure: tree identity mattered, as the species Peronema canescens strongly decreased (by approximately one standard deviation) both canopy openness and understory vegetation cover; whereas tree richness only decreased understory flower density. Further, the smallest plots had the lowest understory flower density and richness, presumably because of lower light availability and colonization rates, respectively. Enrichment influenced herbivorous insects and natural enemies in the understory to a lesser extent: both groups had higher abundances in plots with two enrichment species planted, possibly because higher associated tree mortality created more habitat, while herbivores decreased with increasing tree species richness, in line with the resource concentration hypothesis. Linking relationships in structural equation models showed that the negative association between P. canescens and understory vegetation cover was mediated through canopy openness. Likewise, canopy openness mediated increases in herbivore and pollinator insect abundances. Higher pollinator visitation increased phytometer yield, while impacts of insect herbivores on yield were not apparent. Our results demonstrate that even at an early stage, different levels of ecological restoration influence insect‐mediated ecosystem functions, mainly through canopy openness. These findings suggest that maintaining some canopy gaps while enrichment plots develop may be beneficial for increasing habitat heterogeneity and insect‐mediated ecosystem functions.