How Mean Store Center works

The Mean Store Center function creates a centroid in the mean geographic center of your customer points.

This centroid can be calculated by the following:

Calculating the centroid by number of customers

When the centroid is calculated by the number of customers, each customer point has an equal value. Since the centroid represents a balance point between all customers, it will be located in the center of the customers. If customers are more densely populated on one side, the centroid will be pulled in that direction.

Suppose you want to expand your chain of sporting equipment stores into a new market area. Your existing customer profile shows that you sell to a limited demographic segment: high-income, well-educated people who play golf.

To begin, you might purchase a mailing list of households with similar demographics in the expansion market, geocode them using the Customer Setup wizard, then calculate the centroid by the number of customers. The resulting centroid would be a good place to start looking for a new location.

Calculating the centroid by weighted value

A centroid calculated by a weighted value considers each customer to have an individual value. The centroid is not created in the center of all customers but in the center of the customers who most satisfy the value you've weighted.

Suppose you want to calculate the centroid by customer sales. The location of a customer who has spent $100 at your store will be counted 100 times more than a customer spending only one dollar. When the centroid is calculated, this weighting pulls the centroid toward the more important points.

You can also use this tool to compare the geographic centroid of your customers to your actual store location. Significant distances between the customer center and actual store location can illustrate problems with existing store locations.

Other business uses for centroids

The new optional tool for creating store points based on customer clusters utilizes the k-means algorithm, This is an algorithm to cluster n objects based on attributes into k partitions, k < n. It is similar to the expectation-maximization algorithm for mixtures of Gaussians in that they both attempt to find the centers of natural clusters in the data. It assumes that the object attributes form a vector space. The objective it tries to achieve is to minimize total intracluster variance, or the squared error function.


11/8/2010