A New American Geography: Cultural Ecoregions for Consumer Segmentation

December 11, 2012

I’d like to introduce my first pass at a new American Geography with regional definitions based on common culture, climate and environment rather than political borders.

I started this project because I needed a better way to define different American regions so I can help my clients, who supply or distribute goods in the US, better understand how to “cluster” their store locations and their customers.  Demographic segmentation systems have become a popular tool for analysis of customer data to make retail location decisions; they’re also quite useful in a variety of other marketing strategy campaigns.  I’ve used these data to build sales forecasting models and while they certainly are effective for identifying locations with similar socioeconomic status they don’t usually take into consideration environment/climate or unique cultural attributes that exist in different parts of the country.  This can be critical for certain types of consumer purchase behavior.  For example, if you’re in automotive, you know that consumers on the West Coast and the East Coast are far more likely to buy foreign name plates rather than domestic vehicles.  You also know that 4-wheel drive vehicles are going to sell better in locations with more snow or with uneven topography.  So clustering consumers by income variables doesn’t really capture the many differences in purchase patterns for companies selling cars, car parts or services related to specific vehicle types.  Same goes for home improvement and many other consumer products industries.  Should Home Depot and Lowes stock the same garden supplies at the same time in all parts of the country?  Of course not.

Another problem is the use of political boundaries.  Often the most granular location information for model development is State of residence.  For example, in building replacement rate models for particular car part categories we’ve made model bias adjustments at the State level due to limitations in the available survey data.  This is certainly better than using Census regions or divisions but still falls short when it comes to large states like California, Texas, Florida…even Oregon, Washington, Colorado, Illinois.  The problem is that there often exist clear cultural and/or environmental differences within a particular state that need to be considered.  Let’s say you run a chain of stores that sell tires, like Discount Tire, Les Schwab or even WalMart.  Do you really want to stock all stores in California with the same inventory assortment, fitting the same profile of vehicles?  Something tells me that you’re likely to find a pretty different proportion of Toyota Prius models in Truckee vs Tustin.  Do you think there may be a few more Ford F-150 pickup trucks in Modesto vs Mill Valley?

So, I’ve tried my best to come up with a set of regions that can be used for these types of analytical endeavors and, perhaps, they can be used to help better understand America’s unique cultural landscape.  My colleague and fellow geographer, Steve Scobie, and I started by defining our own set of regions and then comparing – our own little “double blind” study.  From that we came up with some shared boundaries and now I’ve taken it a bit further with some modifications and a set of names to go with each region.  I’ve also created a set of 8 divisions to group the regions when a smaller number of regions is needed.

I would love to get some feedback on this.  I know the Western US fairly well, I’ve lived in the Midwest and I’ve traveled to most parts of the country but there are plenty of areas that I don’t know that well including most of the South and the Northeast.  If you see a boundary that doesn’t make sense to you I would really like to hear what you don’t like and how you might change it.  I would also love to hear thoughts on the region names.  Steve contributed several but I’ll take blame for any that seem incorrect or inappropriate.  My apologies in advance for any place names that someone may find offensive.  I’m sure there are better names out there.  I’m especially interested in names that are already in use by people living in or near the region.  But please don’t submit names with any sort of negative connotation.

Please share your thoughts in the comments section below.  Thanks!

By the way, I had so much fun making this map.  It was produced “old school” by tracing an outline of the “Lower 48” and then using a variety of maps (political, environmental, topographic, demographic) to select boundaries and draw/trace them by hand.  Takes me back to Cartography 101 with the legend of Oregon cartography, Bill Loy; now gone but not forgotten.