Case Study: Geographic Expansion
Problem: Mike owns a successful chain of boutique coffee shops with ~40 locations along the East Coast headquartered in Philadelphia. Mike would like to expand to the Midwest with a regional headquarters in Columbus, Ohio. However, Mike doesn't know the Midwest as intimately as he knows Philadelphia, New York and Boston and is struggling to determine the optimal locations for his new Midwest shops
Parameters: Mike has expressed that he believes his customers are similar to Starbucks customers and that his worst performing existing locations are those where he is in direct competition with Starbucks for business. He also mentions that there were some issues with quality control at an experimental location in Florida. As a result, he would prefer that the first wave of Midwest locations are within a ~3 hour drive of Columbus to avoid the pitfalls of the remote managed Florida shop that has since been shuttered
Parameters: Mike has expressed that he believes his customers are similar to Starbucks customers and that his worst performing existing locations are those where he is in direct competition with Starbucks for business. He also mentions that there were some issues with quality control at an experimental location in Florida. As a result, he would prefer that the first wave of Midwest locations are within a ~3 hour drive of Columbus to avoid the pitfalls of the remote managed Florida shop that has since been shuttered
Step 1 - Limit scope to Midwest zip codes within 200 miles of Columbus, OH
While the below is a helpful visual it still leaves more than 1,000 potential zip codes across the Midwest for Mike to choose from. He's going to need more information to make an informed decision
Step 2 - Perform competitor analysis on Starbucks to better understand customer demographics
Because Mike expressed that his customers are quite similar to those at Starbucks we scraped more than 12,000 Starbucks locations in the U.S. to understand their customer profile. While we could have done this on the roughly 40 existing shops Mike has we felt it would be more statistically significant and geographically diverse to analyze Starbucks data as a proxy. The results indicate that boutique coffee customers are disproportionately high earners with at least a bachelors degree between the age of 25 and 49 years old. Note that blue cells below indicate an overrepresentation and red cells an underrepresentation compared to the average U.S. zip code
Step 3 - Run the demographic algorithm to find the optimal zip codes
The below are Midwest zip codes within 200 miles of Columbus sorted by "Coffee score". The Coffee score is created using a demographic algorithm in order to find zip codes which are most representative of Mike's target customer profile. We have kept the top 10 percent of zip codes as those which may be best for Mike's new shops below. However, in total these zip codes already have more than 200 existing Starbucks locations. The final step is to filter out hyper competitive zip codes which already have multiple Starbucks franchises present
Step 4 - Output the recommended zip codes to examine in more detail
Once zip codes with more than 1 Starbucks are filtered out there are 44 remaining zip codes for Mike to examine in the Midwest. These are geographically diverse with locations near Detroit, Indianapolis, Columbus, Akron, Cleveland and Cincinnati. Even against a massive national competitor like Starbucks, Mike has the opportunity to compete in a new region on favorable terms with the analytics provided by DiGenova Data. This allows business owners who may not have a data science team in house to focus on their products and expansion efforts without wasting precious time deliberating on where to go or guessing where their best customers are