I just had my 13th anniversary at work a couple of weeks ago. I am always amazed at how many people keep up with that kind of trivia. I got emails, texts, and phone calls from quite a few people….many that I haven’t talked to since this time last year. Regardless, it has been a time to reminisce and talk about the good ole days. I remember, in my first few months at work I quickly realized that we needed to analyze our store performance differently. We basically looked at all 4,000+ stores as if they were exactly the same. Within chaos lies opportunity…right? I went to work on figuring out how to group stores with similar sales patterns, unit velocities and/or brand preferences.
I teamed up with our analytics team and after a lot of trial and error, data scrubbing, analysis, etc. we landed on a process and eventually had a tool built. The first clustering run took me a full week to run through the process as defined by hand. The clustering tool moved that time to within a couple of hours.
At this time, I’m going to pause the story and share that last Wednesday night Prof Ames described the almost identical process and math to us in less than an hour. I was blown away and more than a little humbled. Now to continue with the story….
The first category where we implemented store clusters was a small $600 million category that was running -6% year over year sales. The category had a new buyer and he was desperate for help. We implemented a 5 cluster set, created with a k-means algorithm and within weeks was running +10% sales growth on less inventory. Needless to say I was very excited and soon had more work than I could handle. This department now has a hundred or more analysts doing category analytics, including clustering….it is required.
The funniest thing was one night as I arrived home from work my young, 5 or so year old, daughter comes running out to me and asks….”Did you get to cluster something today daddy?” She thinks she is so funny! Of course I did!!
My Analytics Evolution 