In order to produce the Consumer Price Index, the Bureau of Labor Statistics collects its sample frame using a very narrow definition of the target sample unit - a retail establishment or 'outlet'. Specifically, an outlet is defined by the unique combination of operating name, mode (e.g. internet, brick-and-mortar), and if brick-and-mortar, the exact physical address of the store. While it is necessary to determine a precise location for pricing purposes, this definition of an outlet may not be ideal for sampling. This paper examines whether or not the definition of an outlet can be broadened in order to simplify data collection and to allocate the sample more efficiently, without introducing bias or nonsampling error. Specifically, the effect of both location and franchise status on price change is modeled and evaluated to determine which is more relevant in defining an outlet.