Changes in temperature affect consumer-resource interactions which underpin the functioning of ecosystems. However, existing studies report contrasting predictions regarding the impacts of warming on biological rates and community dynamics. To improve prediction accuracy and comparability, we develop a framework that combines two approaches: sensitivity analysis and aggregate parameters. The former determines which biological parameters impact the community most strongly. The use of aggregate parameters (i.e., maximal energetic efficiency, ρ, and interaction strength, κ), that combine multiple biological parameters, increases explanatory power and reduces the complexity of theoretical analyses. We illustrate the framework using empirically-derived thermal dependence curves of biological rates and applying it to consumer-resource biomass ratio and community stability. Based on our analyses, we present four predictions: 1) resource growth rate regulates biomass distributions at mild temperatures, 2) interaction strength alone determines the thermal boundaries of the community, 3) warming destabilises dynamics at low and mild temperatures only, 4) interactions strength must decrease faster than maximal energetic efficiency for warming to stabilise dynamics. We argue that directly measuring the aggregate parameters should increase the accuracy of predictions on warming impacts on food webs and promote cross-system comparisons.