Learning curve research has found that rates of learning can vary across similar settings, such that cumulative experience is a necessary but insufficient predictor of learning curve slope. One explanation for this is that how the learning process is managed affects rates of learning. We investigate an additional possibility: by pursuing two dimensions of performance improvement simultaneously, effort invested in one may inhibit learning rate in the other. Using a sample of sixteen academic and community hospitals adopting a new surgical technology, we demonstrate a tradeoff in rates of learning on two dimensions-efficiency and technical difficulty-providing support for our proposed explanation of learning curve heterogeneity.