Peter S. Lindquist, Ph.D.
William A. Muraco, Ph.D.

CME offers a unique integrated package SALES (Segmentation Analysis-Location Evaluation System) that combines the latest market segmentation data mining technologies within the framework of a spatially based geographic information systems problem solving approach. The package represents a significant extension of standard market segmentation analysis, by incorporating a full range of business geographic service capabilities.  These unique approaches coupled with the latest methodologies in market segmentation analysis generate a geographic based sales modeling approach that defines the distribution of potential buyer segments of specific products and services over multiple scales ranging from as small as a census block group to the national level. Through the use of a Geographic Information System (GIS), sales potential is mapped to display the distribution of demand within an area of interest.  Sales potential is then applied in a GIS-based location analysis system to designate optimal retail/service locations that will maximize the accessibility of firms to their market segments.

The initial analysis uses quantitative techniques with no a-priori assumptions regarding key imagery segments present in the primary survey data. The analysis is used to build unbiased segments based on the importance that spatially referenced consumers place on specific services and product attributes associated with a gasoline station. This multi-stage analysis having no a-priori assumptions regarding respondent differentiation is generated initially to define the potential buyer clusters unique to the industry in question.  The analysis is then coupled with quantitative approaches having a theoretical foundation to validate the level of differentiation between segments and to identify the key “drivers” that are most responsible for the segments.  Finally, interaction detection modeling is used to profile the primary demographic, social, lifestyle and behavioral characteristics associated with each segment defined by the earlier buyer segmentation analysis. The interaction detection modeling provides the foundation for linking the respondent level data to geographic space.

In order to provide an illustration of these segments, the Cash No Frills Cash buyer segment is used as a case example to illustrate the potential sales surfacing and location analysis components of the SALES technology.


Stage 2 -- Geographic Referenced Market Segments

 

In the previous segmentation analysis phase the “No Frills, Cash Buyer” segment was associated with the following characteristics:

•  predominantly white population;

•  middle income ($40,000-$50,000 annual household income);

•  “some college” in terms of education.

 

 

Using this GIS-based approach, it is possible to focus on smaller areas of the map very quickly to study regional markets.  In this case, Northeast Ohio, featuring the Cleveland and Akron metro areas, is illustrated in the following map:

At this stage of the work it is critical to relate the market segments identified in the survey research to the socio-economic, family status and ethnic/cultural characteristics of the statewide population in order to identify where each of the market segements are most likely to occur.  This process would be repeated for each of the six segments identified in Stage I.  The maps resulting from this procedure are useful in showing where block groups containing this market segment are likely to be located.  However, they do not effectively show the concentration or potential sales volume of this segment of the market population among different locations over the state.  Some areas will have higher population densities of this market segment in more compact block groups clustered within individual areas.  Others will not.  It is therefore advantageous to map the distribution of the sales potential concentrations of each market segment.  Knowledge of the proximity of these market concentrations to existing or potential stations and c-stores can be critical to the success of those enterprises.  The next stage of this application shows market concentration maps generated for this market segment over the State of Ohio.

This final phase of this application focuses on mapping the sales potential of this market segment. Specialized software developed as part of SALES uses the block group data produced in Stage 2 to map the geographic proximity of this market segment to all areas in Ohio. The following statewide map is quite different from that produced in the previous stage.  In this map, all areas in the state are mapped with respect to the surrounding concentration of this segment of the market.  These concentrations are expressed as a percentage of the highest market concentrations found in the state (in the Cleveland Metro Area).  These may also be expressed in terms of potential sales volume rather than population by incorporating data for spending habits of this market segment. 
 

It is clearly evident from this map that the highest concentrations are found in the suburbs of the major metropolitan areas of the state, with pronounced “corridors” extending from Cincinnati through Dayton to Columbus, and from Cleveland to Akron. 

The Northeast Ohio region is again illustrated in the next map to examine more closely the variability in the concentration of this market around the suburban areas of Cleveland and Akron.  The advantage of using a geographic information system in this application is that state-wide analyses can be run with very large volumes of data as a means to review wide areas; however, localized trends within individual regions can be examined more closely where needed.  Decisionmakers thus have the ability to carry out a detailed examination of market segment concentrations over very wide areas and decide quickly upon which areas to focus.

A more critical factor in assessing the potential location is to identify market opportunities relative to the competition.  In other words, the concentration of the segment in geographic space alone may not represent the true potential market since it ignores areas that may already have a heavy concentration of competition (saturated market). In the following map we extend the market potential concept to identify the segment’s concentration relative to competitors operating within the regional market.  It is not surprising to find that the competitor potential map appears similar to the market segment potential map.  The segment’s concentration, relative to locations of the competitors, remains concentrated in the core urban axes between Cleveland and Akron.
 

A final analytical step incorporates location-allocation modeling to identify the ideal pattern of all regional station and c-store locations, and in turn can place these facilities within close proximity to this segment of the market.  The map below shows existing station and c-store locations as crosses on the map; the dark circles attached to the lines represent the shift in station locations to this set of “better” locations. In some cases, stations barely shift at all, while others shift significantly.

The box marking a section of the East Side of Cleveland in the above map provides a close-up of a station relocation seen in the next figure.   The figure shows one proposed station shift from its present location (designated by a cross) along St. Clair Avenue in Cleveland to Green Road (designated by a closed circle) at the Cleveland corporate limit. 

A survey of surrounding block groups within a two mile radius of each station location shows that the new location is in closer proximity to the “No Frills, Cash Buyer” segment of the market described in this discussion.  The following table shows the differences in the primary demographic, social, lifestyle and behavioral characteristics associated with this segment. In all categories, the new location shows an improvement in location relative to this segment, especially with regard to the education and income characteristics.

Further analysis over the entire study region shows that the model produced a 16.6% increase in the market population of this segment within a two-mile radius of new stations in contrast to the old.  While this illustration shows how location-allocation models can be applied in this system with respect to one segment of the market, it should be noted that any location-allocation analysis to identify ideal station locations should include all market segments to assure the widest possible customer base.