Abstract There is increasing recognition that pollination deficits are limiting crop yields world‐wide. However, management strategies for optimal insect pollination are still unclear for most crops. Current management focuses on providing high densities of honeybees, but recommended densities are highly variable, even within single crops and cultivars. We performed an extensive literature search to record honeybee densities (colony density and/or flower visitation rates) and crop productivity (fruit set, seed set, fruit weight and/or yield). Effect sizes were represented as the difference in crop productivity between the two most extreme levels of honeybee densities. Out of 795 reviewed studies, only 22 analysed the effect of at least two levels of honeybee densities on crop productivity (reporting 60 resulting effect sizes in total). Moreover, few recommendations for crop pollination management are based on results from controlled experimental designs, and with comparable methodology. We found that both colony density and visitation rates increased all the productivity variables. However, effects were nonlinear for visitation rates, suggesting that there is an optimum (mean of 8–10 visits per flower) beyond which more honeybees are not beneficial (and can even be detrimental) for crop productivity. Effect sizes for visitation rates were greater than that for colony densities, suggesting that visitation rates are a more direct measure of the pollination process. Data on the relation between colony density and visitation rates are lacking. Interestingly, effect sizes for visitation rates were greater for crops with separate sexes than those with hermaphroditic flowers; therefore, the benefits of honeybee pollination vary according to the crop biology. Synthesis and applications. Current practices for crop pollination assume that more honeybees are always better for crop yield, even if the effect of this management on crop production still unclear. In contrast, our analyses suggest that there is an optimum for honeybee densities. Despite the importance of honeybees and pollinator‐dependent crops world‐wide, there is a lack of studies designed for finding such an optimal level of crop pollination. Our analyses further suggest that visitation rates could be used as a proxy to guide management recommendations such as colony density and spatial arrangement.