Process improvement takes place in the context of ongoing work, so people usually face the dual pressure to produce output and to build capability. Repenning and Stermanís (2002) study of two process improvement initiatives developed a causal loop diagram characterizing first-order improvements which boost output from existing processes and second-order improvements which enhance the capability of underlying processes, but the study stopped short of building a simulating model. This paper starts from the feedback structure they present and constructs a system dynamics model that formalizes the critical interaction between first- and second-order improvements as options for governing production. Analytical results characterize the optimal tradeoff between working harder and working smarter. However, practitioners generally must manage this tradeoff lacking adequate knowledge of the parameter space to find the optimum. Results demonstrate tipping points in the dynamics of process improvement, identify perverse behaviors that are likely to thwart the good intentions of practitioners, and show how the feedback structure of process improvement presents challenges to agents facing the dual pressures to produce and improve. By moving from causal loops to a simulating model, the paper also provides an example of how formal modeling yields more nuanced understanding.