Literature on location strategy contains guidance on what developers should look for when evaluating the suitability of a property or market. For example, daily traffic count should be at least 25,000. Frontage road should be undivided. Site should be highly visible in both directions of traffic. No more than a few competitors. Income and population density should be robust. And so forth.

Historically, these variables have been studied, weighted, and included in models that car wash equipment suppliers and developers use to define store trade areas.

The accuracy of these models is based on the principal of similarity.

By finding sites that have similar attributes, it follows that a good estimate of store performance would be an average of the subset.

One problem with this approach is bias. Bias results because the similar sites used to calibrate the models are stores with above average performance rather than average or below average performance. Figuratively, the predictive models observe more homeruns than doubles or triples and extrapolate this in the results.

There are a number of ways to prevent bias from skewing results. This begins with purpose. For example, car wash equipment suppliers offer their clients site evaluation services to help them understand the business before they become fully invested.

The results from a site evaluation are crucial for short-term planning such as developing a cash budget. Components of a budget are time period, desired cash position, and estimated sales and expenses for the period.

However, site evaluation services usually do not have the scope for long-term planning such as an exit strategy. For example, classical location analysis procedures are usually limited to assessment of market and local competition.

Here, the results are used to identify stand-alone value and understand potential performance of a specific site (property). Absent from this approach is data that can be used for decision making related to total shoppers and sales flow.

Sales flow is the calculation of total potential sales originating from an area. Total potential sales are used to determine consumer purchasing power in an area and then compare it with nearby areas to gain perspective of how attractive one area is in relation to another.

Here, the gravity model has been a proven model in the location-analysis industry. A gravity model calculates total potential sales as a function of population in an area, mean per capita income, and distance to adjacent areas.

This method is used extensively by convenience retailers such as grocery stores, convenience stores, gas stations, and other support services where distance is a crucial factor in consumers choosing one store over another.

Although the gravity model is widely used, researchers find it has little sensitivity to demographic variations or market segmentation. Consequently, location analysts have augmented such modeling with the use of geographic information systems (GIS). GIS is a framework that provides the ability to capture and analyze spatial and geographic data and allows for a more granular approach to location analysis.

Shown above is an illustration of a raster analysis.

Raster datasets represent geographic features by dividing an area into rectangular cells laid out in a grid. Each cell has a value that is used to represent characteristics of that location such as population, income, tenure, etc.

Generally speaking, location analysis with GIS involves three techniques: Raster calculations are used to derive potential market; polygon analysis is used to help determine store trading areas; and a gravity model is used to show market potential for each relevant area.

Accuracy can be improved by combining GIS, multiple analysis techniques, and industry knowledge whereas precision can be improved by avoiding bias. This can be achieved by using dummy data and comparing the results with real-world circumstances.

Bob Roman is president of RJR Enterprises — Consulting Services ( You can reach Bob via e-mail at