Not Your Father's Trade Area ... Beyond Block Group and Census Tract

Since the 1990’s, retail real estate analysts have fallen back on the same ancient methodologies to tackle trade area delineation. Unlike your father’s vintage Chevrolet, the half century old polygon based trade area does not appreciate with age. Sure, there is something that can be said about a “time-tested” model, but retail above all other industries should value an organization’s ability to adopt new technologies and embrace change.

Today, marketing strategists and GIS professionals incorporate the latest spatial analytics platforms to churn terabytes of data and render retail trade areas in a timely manner. Trade area analysis has quickly matured as a key instrument in any real estate manager, market planner or real estate strategist’s toolbox for its wide range of uses. With incredible strides in data mining, consumer research and demographic data, retailers and analysts are developing new approaches to predict consumer spending habits and forecast behavior. With or without customer data, a new trade area methodology must take shape to account for these advancements.

Who Uses Trade Areas within the Organization?
Retailers use trade areas primarily because they need a replicable, defensible, and systematic way of describing the origins of customer traffic to their stores. Without trade areas, there would be no practical way to compare and evaluate locations based on the surrounding factors that drive store sales. Trade areas can be as simple as a radius, or they can be much more elaborate, utilizing the sophisticated and powerful spatial modeling techniques that have come of age in the last decade.

In the context of retail companies, trade areas (also known as trade zones or catchments) have several applications for the positioning, forecasting and optimizing of store locations. Senior Leadership in the areas of Real Estate, Marketing and Operations leverage trade areas to aggregate demographic information, delineate optimal spacing for stores, observe impact to sales from changes in competition or from the influences of advertising campaigns, and aid in the calculation of sales forecasts and profitability.

However, more specifically for the Real Estate Manager and Market Planner, accurate trade area delineation is useful when visualizing the spatial distribution of customers and sales, as well as to identifying gaps in the market.  Accurate trade areas also allow for more precise forecasts of cannibalization allowing the optimal spacing of stores to be calculated without sacrificing sales.     

Once tasked with the responsibility of fulfilling the development strategies laid out by the company, the real estate team must identify good locations fast. When the CEO of a major retailer promises Wall Street that they will open one thousand locations in the next three years, real estate managers and market planning teams find themselves in the unenviable position of delivering deals quickly, and the ability to rapidly forecast sales and construct prioritized development strategies is paramount not only to their success but also to the investors success as well.

Market Planners and Real Estate Strategists also use trade areas for competitive research. Some large retail brands view their store development strategy as a way to protect their markets from competitive intrusion. The point of view here is to build out a market to leave as little opportunity for the competition as possible.

Further, marketing departments will analyze customer segmentation and demographics within a store’s trade area to develop a marketing strategy that drives customers to stores, or if closing a store they will construct a strategy to drive existing customers to the closest remaining locations. By knowing more about the customers within a store’s trade area the marketing strategy can become more customer focused. Merchandising, pricing as well as traditional and direct marketing campaigns can be driven from this information.

What Questions Do Trade Areas Answer?
Recognizing the importance of where customers live and what boundaries customers face in reaching a given store is the first step to reaching a successful real estate strategy. Any practical trade area should visually depict where the assessed store’s consumers reside and the physical boundaries that constrain the store from dominating outside its core markets. Within each trade area, census data and third party data can be included to gain a better understanding of the geospatial makeup of potential customers. If customer profiling is used, future store trade areas can be sought that match the most profitable customer demographics.

Another instrumental use of trade areas is to observe how competition, barriers to movement and settlement patterns manipulate the trade area. Understanding the relationships between these variables and a store’s trade area boundary can be extremely beneficial in the planning of a new store location. As competitors move into a market trade area definitions can be used to depict the change in customer patronage and thus the impact on trade area draw, due to the change/addition of competition.

By far, the most important application of trade areas to retail and other location centric companies is in its ability to forecast future sales.  Through the accurate deduction of a trade area model, analysts can deploy that trade area model to forecast EBITDA, cannibalization and ROI of new stores, relocations and renovations. With today’s volatile economic environment, optimizing capital spend is vital in increasing or maintaining a concept’s profitability.

What are the Classic Trade Area Methodologies?

1. Circular Trade Area - Defined as a circle using a simple pre-defined radius, the circular trade area is the easiest, quickest and least expensive methodology used. However, this methodology does not take into account geographical barriers (such as rivers and mountains) or competition. Depending on which predefined radius is used, adjacent stores may or may not overlap. In short, circular trade areas often result in an incorrect delineation of trade area due to the simplicity of the methodology.
2. Percentage of Customers -  One of the most popular techniques used today (when customer data is available), is the calculation of the percentage of sales as contributed by customers to depict a trade area. The boundaries of the trade area are drawn using a convex hull that contains 60-80% of customers or customer sales. These trade areas are more precise than circular trade areas, but they are descriptive, not predictive: when a new site is being evaluated, there is no sales distribution to capture. Thus, these trade areas are a step on the way to a more predictive trade area model.
3. Travel Distance/Drive Time Analyses - This approach uses driving distance or drive time intervals to draw the trade area boundary. Unlike the circular trade area and the percentage of customers method, drive time analyses have a tendency to depict irregular shapes as they reflect the impact of the road network on trade area extent and customer patronage of a particular store. Drive times are a definite improvement over circular trade areas, but like circular trade areas, they do not consider the competitive landscape of the market.
4. Theissen Polygon - The Theissen Polygon and Voronoi Diagram are geometric techniques for delimiting theoretical trade areas for a number of stores. This methodology assumes that all of the stores taken into account are similar in size and market products for similar price; consumers will only purchase products from the closest store. Compared to above mentioned methodologies, the most important feature of this method is in its requirement that all trade areas are non-overlapping. Recent advances in GIS techniques have improved the Theissen Polygon to improve its flexibility; namely the Weighted Voronoi Diagram and Network Voronoi Diagram. Theissen Polygons do not work particularly well for concepts where customers are free to shop a variety of stores based on convenience. They are, however, especially useful for managing territories in concepts where the retailer delivers merchandise to the customer (i.e. pizza delivery routed through a call center), as they maximize the efficiency of the delivery process by assigning the closest store.

Other Classic Methodologies

• Reilly’s Law - Developed in the early 20th century, Reilly’s Law of Retail Gravitation merely adds a second variable to the Theissen Polygon methodology. While Reilly’s Law was traditionally formulated to approximate the pull  on an area between two cities it was later retrofitted to account for a retail site’s attractiveness and produce a Breaking Point that can be used to draw a boundary between the two sites. Reilly’s Law is most commonly used today in approximating trade area breakpoints between towns in sparsely populated (i.e. Great Plains) regions.

• The Huff Model - In response to his criticisms of Reilly’s Law of Retail Gravitation, David L. Huff proposed a statistically-calculated probabilistic method termed the Huff Model. In order to create a trade area that would account for change in a store’s size, cannibalization and competition Huff hypothesized a method that could generate customer volume estimates for new and existing stores, map the probability surface of a customer traveling to a given store, and estimate sales potential for each observed store location. Due to the complexity of the Huff Model, this model requires a lot of data and repetition to work correctly. Like all gravity models, the Huff model is appropriate in situations where the competitive network is well-defined, the concept is convenience-driven, and retail potential is clearly articulated.

Where P is the probability of a customer going from origin i to store j, A is a measure of attractiveness of store j, D is distance from i to j, and lamdba is a distance decay exponent that reduces propensity as distance increases.

A New Approach
The approach of Tango Management Consulting increases accuracy and includes more location specific variables at greater levels of detail made possible by advances in spatial data processing. Some of these variables and approaches are new, some are client specific while other variables are improved by advances in drive time estimation, use of block census geographies, and fast and accurate geo-coding (assigning customers and competitors an accurate location) techniques.

Getting Granular
With the advent of more effective methods to gather and store customer information, Marketers and Real Estate Strategists have greater opportunities to learn more about their customer base. Specifically regarding real estate, increased granularity in customer data leads to better sales predictions and, thus, better site decisions.

Granularity also allows retailers to reorient their thinking around how to model customer behavior. In the past, retail analysis was store-centric, with the store as the center of the universe: open the doors and the customers will come. Instead, in today’s multi-channel retail environment, retail modeling must become customer-centric. As a result, the emphasis in trade area analysis has shifted to modeling customer behavior as precisely as possible, given the myriad of choices available to today’s customers.

Aggregating block groups, or even worse census tracts, to build market penetration trade areas simply does not allow GIS analysts to get close enough to the data. When each block group could be inhabited by anywhere from 300 to 37,452 people according to the 2010 U.S. Census, appending customer specific data to a block group could be misleading. What may be consistent with 1/3 of the block group might not be representative of the entire block group. That is why it is imperative that GIS and Market Planning experts think smaller. With today’s powerful computer hardware and innovative statistical applications, it is possible to go one step further; block level.

Previously, the spatial analysis of data by block level was considered too computationally intensive to be practical. Since blocks contain a dozen or so households instead of hundreds of people, the size of the data set is exponentially multiplied complicating statistical analysis. Leveraging state of the art tools and applications, this revolutionary approach is beneficial in trade area delineation because it lets analysts segregate households by demographic and transactional data. Furthermore, household specific variables can be appended to each block vastly increasing accuracy and painting a clearer picture of the dispersion of customers and customer sales over a given area.

This more granular data can then be used to construct trade area models that are multi-dimensional in nature. Where in the past a trade area might be described as a 3-mile radius, today’s trade area models incorporate drive time, inbound/outbound orientation, competitive intensity, demographic segmentation, and precise geographic data to arrive at trade areas that are much more accurate and predictive than in the past. This increased accuracy allows not only for more accurate forecasts into the future but for more reliable marketing and market planning activities within the current store portfolio. For the Retail Real Estate Analyst, this new approach is highly beneficial for a number of reasons:

1. More accurate boundary - Retail trade area boundaries can be drawn to precision to include higher percentage of in-profile customers due to the block level’s smaller size compared to block groups. This is especially useful in concepts that have very small trade areas.
2. New Options for displaying trade zones - By aggregating households into smaller data sets, demographic data is more homogeneous with their corresponding data sets. This is especially important in urban areas where a block group may include several corners of neighborhoods with dissimilar demographic data. Using spiders and hot spot maps, we can convey more information than a simple polygon.
3. Factors in additional information including demographic data, barriers, drive-time, sales per household, retail density, store density, shop up/shop down, sales per effective households and competitive density - Similar to demographic information, smaller data sets are also much more representative of other variables like Household Sales, Retail Density and local market penetration. This data can be aggregated from various published data sources using advanced GIS techniques.
4. More precise sales forecast modeling - Since trade area delineation leveraging block level data creates a tighter and more precise trade area, demographic information can be exported more precisely. Sales forecast modeling, which utilizes these variables as independent variables, can get closer than ever before to predicting store sales.
5. Based on rules, we can automatically predict precise trade areas for new stores - Where most approaches in the past were simplistic in comparison, today’s forecasted trade areas come much closer to describing the actual origins of customers in the future. This results not only in better real estate decisions, but improved marketing as well.

Conclusion
As retailers continue to adapt to changes in customer behavior and demand, it is imperative that successful retail brands leverage the latest cutting edge tools to help them adapt to the rapidly changing retail environment. The days of the black box model are over, and your father’s trade area will not cut it in this ultra-competitive environment. The use of multi-dimensional techniques and the delineation of trade areas at the block level provide an output with more detail, more data and better results.

Reprinted with permission, Tango Management Consulting, LLC