Abstract: Households adopting new agricultural technologies often face labor constraints influencing the extent to which such technologies are productive and profitable. Such labor constraints differ for nuclear and extended-family households. In a randomized control trial, we estimate the heterogeneous treatment effect of an efficacious fertilization technique called microdosing by differences in household structure. The encouragement design which allocated starter packs and microdosing training to assigned households induced extended family households to reduce labor to agricultural activities, while nuclear households increased such labor activities. These differentiated effects are dominated by households who had previously used fertilizer. Thus, microdosing does not completely relieve the binding labor constraint for extended households who previously used broadcast fertilizer methods. Although nontrivial labor allocation is necessary for both broadcast and microdosing, for extended households, microdosing lowers total person-days to agricultural production by 18% relative to mean labor allocation at baseline, whereas nuclear households increase labor allocation to agricultural production by 36%.