Synthetic chemical pesticides can enhance crop yields but also have undesired effects. Alternative 'botanical insecticides' may also have non-target effects on pollinators and biocontrol services. Employing action thresholds (ATs) can reduce pesticide (whether synthetic or botanical) use compared to fixed-interval applications. Here the azadirachtin-based botanical formulation NeemAzal and a neem seed extract (NSE) were evaluated in field spraying trials alongside commonly-used synthetics (Voliam Flexi [chlorentraniliprole plus thiamethoxam] and imidacloprid) in developing ATs for the regular and cosmopolitan cauliflower pests Brevicoryne brassicae, Plutella xylostella and Spodoptera litura. We considered the size of the S. litura larvae infesting the crop in order to derive ATs. ATs per plant were higher for NeemAzal (0.55 larvae for P. xylostella and 3 larvae for large-sized S. litura) than for Voliam Flexi (0.30 larvae for P. xylostella and 0.80 larvae for S. litura) but were similar for B. brassicae (50 individuals). Higher ATs when using azadirachtin were associated with the diverse modes of action of botanicals, for instance NeemAzal and NSE deterred oviposition of S. litura. Although the exact values of ATs are likely to have regional limits, our approach can be applied for determining ATs against common lepidopteran and aphid pests in many other vegetable crop agro-ecosystems.