Years ago I was trying to sell site evaluation software to the commercial lending group at Freddie Mac. It had 15 underwriters around the country with huge piles of loan requests on their desks and very little time to analyze each deal. The person I was working with described his problem like this:
“There are only about 10 criteria we need to evaluate in order to approve or reject the loan. Unfortunately the 10 criteria are usually different for each deal!”
The collateral for the loan might be an apartment building, a convenience store, a strip shopping center, an office building or a warehouse. It could be in an urban area, in a small town, near a college, in an affluent area, leased to a credit tenant, have no parking or located in a prime retail district. I wouldn’t want to have to design the “standard checklist” for underwriting these loans!
Many chain store operators use a “checklist” approach to evaluating potential sites for new stores. They don’t have the same challenge as the lenders at Freddie Mac because they are usually considering a site for a very specific use. RadioShack can always assume that it will be selling consumer electronics from a fairly small retail space. Dunkin’ Brands has a few different prototypes, but it is looking for high visibility locations to capture frequent and transient coffee and snack customers. However, there are still many different settings for a RadioShack or Dunkin’ Donuts store. Population density, employment, competition, visibility, cotenants and many other variables create a unique “context” for each site, requiring a different set of questions in order to understand its sales potential.
How many people do we need within a three-mile radius? It depends. Will the resident population be the only source of customers or will the daytime population contribute to sales as well?
How much competition is too much? It depends. Is there enough retail in the area to create a regional draw that will increase the effective number of customers in the area for everyone to share?
What is the minimum number of parking spaces needed to ensure that customers will come to the store? It depends. Is it an area where people have several stores to choose from with ample parking or is it an urban area where people expect parking to be a headache and will not be discouraged if the number of spaces is fairly low (or even zero in dense urban areas)?
If we build a statistical model based on key drivers of sales such as population, income, site characteristics and competitive presence, how many different models do we need to account for the different contexts? I’ve seen cases where modeling vendors had almost as many models as existing stores. If we try to create groups of stores with similar contexts, how much variation between the stores in a group can we allow before we’re not really comparing apples to apples?
Sorry if this makes your head hurt. It’s been making my head hurt for nearly 20 years. I suppose I could quit worrying about it and find a simpler business. But now it’s become an obsession and I know that many others share this affliction as well. So, let’s press on…
Back to this idea of context. It’s something that our brains handle quite elegantly. With a little experience we can quickly zero in on the features that are central to the identity of a site. We can recall similar sites and probably remember whether the sales were great, horrible or just okay.
If we talk about what we are seeing with other experts and take notes, we might see some patterns over time. These patterns can be used to label existing stores and match their features to new sites that are being considered. A computer can be “trained” to compare the patterns between the existing stores and new stores and the sales from the existing stores can be used as a guideline for estimating the sales of a proposed site.
Imagine this type of framework for evaluating sites combined with feeds from social media and a highly interactive research tool that could be used by anyone who can open a Web browser.
This is the future of site evaluation. Hopefully we can all work together to further develop this vision and begin building it into decision processes!