This past May at the IIE National Conference, we presented a case study about the challenges and benefits of simulating retail store operations. Simulation is a tool that utilizes a computerized model of a real world system or process that is run over time periods reflective of reality. It is a great tool for a company to test changes to its operations in a virtual environment, without incurring the cost, headache and damage that could be caused by testing a change live. As with any kind of modeling, the model is only as good as the data used. For many years, simulation has been used in industries such as Manufacturing, Distribution and Transportation. In some ways, it can be less complex to apply simulation to these types of industries, as many of the processes happen within a controlled environment. One of the main challenges in applying simulation to retail is the introduction of the customer into the model. Companies can control machines, assemblies of products and other processes that happen in their own plants. However, retailers cannot control how customers move through their stores and interact with their employees and products. Customers pose one of the greatest challenges to a retail simulation model. How do you model human behavior in a virtual setting? The first and best scenario is to collect any data that the retailer already has available in their in-store systems. System-wide data for an entire chain will always prove to be the most accurate and effective. Any data points that the system doesn’t provide can be obtained through observations, interviews or estimates. The best manual method is to gather actual observations from the store. With more observations across more stores, the data becomes more credible. While interviews and estimates can get the model directionally correct, they are the least preferred and accurate method of modeling the customer. As you consider modeling an environment where customers interact with your company, make sure to consider the complexity that a customer introduces to the model. Think through the parameters above as the first steps to determine what kind of data you need to model customers in your simulation.