Thursday, September 18, 2008

Bye-Bye 99 Cents Only Stores
(99 Cent by Andreas Gursky. Chromagenic Color Print. 1999)

They are apparently giving up on Texas. Why should I care? Well, my first big paper at B-school was written about them. That paper pushed me down the path that got me the job I have now by exposing me to the power of statistical analysis.

Our team's goal was to recommend ways to promote new stores. Now before we could get there, I wanted to see what it was that brought people into a 99 Cents Only store. We couldn't easily poll people, so we thought about using demographic information from American Fact Finder. We had some crude data from the company about which were the best performing stores in Houston, the worst, and the middle. I got demographic information for every zip code in which a store was located and every adjacent zipcode. At the time we did this analysis, I only vaguely knew what a regression was, and certainly did not know what a logistic regression was, which would have been quite useful. Instead, I just looked at the correlation of various demographic features with store success.

This is where the magic came in. Of all the demographic pieces of info we looked at--race, income, family size, marital status, unemployment level, etc.--by far the most highly correlated one was percentage of renters versus home owners. This fit right into what we had learned in marketing--that behavior is a better predictor of preference than demography. It also gave us a perfect way to target promotions--circulars sent to apartments near 99 Cent Only stores.

(And for the record, 99 Cents Only stores were pretty decent stores for what they were.)

Labels: , ,


Post a Comment

Subscribe to Post Comments [Atom]

<< Home