A primary activity in operations research and the management sciences is the creation of quantitative models to support decision making. They can be optimization models, simulators, or machine learning tools fitted to available data. Often, the sophistication of the modeling exercise forces analysts to create complex architectures. Tools for explainability and interpretability become essential to reinforce stakeholders’ trust in the model forecasts and increase transparency. This tutorial reviews the role of sensitivity analysis as a toolset for interpreting and communicating model results. We examine the evolution of sensitivity analysis in the management sciences and analyze available methods through the lens of four primary analysis goals: factor prioritization (or feature importance), trend determination, interaction quantification, and stability (robustness). We present a variety of techniques, starting with local methods and moving on to global methods, and critically evaluate the insights they provide in relation to the aforementioned goals. To illustrate these methods, we use a classic example: the optimal lot-sizing model developed by Harris in 1913. We conclude with a series of steps that analysts can follow to select the most appropriate sensitivity method for the analysis at hand.