This dissertation comprises three essays in which we develop optimization, econometric, and simulation models to help traditional retailers improve in-store operations. Our modeling efforts aim to tackle inventory record inaccuracy (IRI) and suboptimal staffing levels, both of which are pervasive problems in retailing and cause non-trivial profit loss. In the first essay, we devise two optimization models that represent current practices in industry to minimize costs induced by IRI: daily-fraction and all-or-none inspection. We further perform a case study to identify deficiencies of store operating practices given different risk preferences. Our findings provide practical guidelines for managers to design cost-efficient inspection policy. In the second essay, we develop a dynamic simulation model to analyze multiple antecedents of IRI. Based on simulation results, we derive two hypotheses on the association between IRI and labor. The panel data analysis shows that both the level and the mix of store labor have strong impacts on IRI. Our analysis derives qualitative insights for retail managers to prevent the occurrence of IRI. Finally, in the third essay, we perform an empirical study to improve staffing decisions in retailing. We first develop a response function to quantify the impact of labor and traffic on sales. Grounded on the function we propose a traffic-based staffing heuristic, which performs closely to the optimal and outperforms existing staffing levels in counterfactual experiments. A major contribution of our study is to quantify the benefits of delivering labor plans based on traffic information. Also, the staffing approach is easy to use and saves the need for traffic forecasting.