Shopping Basket based business planning...

My last post was on the creation of a "Customer Relationship Manager" position as a first step towards customer centricity. I had intended to continue on this topic. But last week the local Spencer Retail outlet shutdown in our residential locality. I was thinking over why it did not work out for the retail chain to continue with the store. So I have pushed the CRM position story to the next post.

We had moved into this locality about 7 years back. At that time, in the place where the current Spencer Retail was located, a local entrepreneur had started a supermarket format called Sunrise Supermarket. Since this was the only super market around and also it was closer to our residence, we started frequenting it for all our household purchase. In fact soon it was a Sunday ritual to drop into Sunrise on our way back from church and gorge on some puff cakes.

All our shopping trips started and ended with one trip to Sunrise. Slowly we adjusted our weekend activities to be able to visit Sunrise when the crowd was at its lowest .. which was saturday afternoons.

A couple of years back, Sunrise wound up and gave its space to Spencer Retail. A D*Mart outlet opened up about a mile further down. While Spencer tooks its own time to start operations, D*Mart was quick to setup and beat Spencer to its opening. My wife and myself started frequenting D*Mart. However, we always tended to stop at Spencers on our way back from D*Mart. D*Mart did not stock non-veg items and we always needed to pick up chicken and bacon from Spencer.

Soon wife decided to drop D*Mart and visit only Spencer. There were two primary reasons for this -- (1) She did not like the quality at D*Mart, and (2) We were spending too much time at the check out counter at D*Mart. I probably know the reason for the second one... D*Mart was cheaper than Spencer and so attracted maximum crowds in a two store locality.

But everytime we came back from Spencer we were getting more and more frustrated. The number of items on our shopping list that were not available in Spencer was increasing every week. So after our trip to Spencer, my wife still insisted on visiting the local market for stuff that we could not get at the Spencer outlet.

In fact, I remember about an year back, my wife was so frustrated that she actually called for the store manager and complained about items that were no longer available. She told this person that at this rate they will soon have to shut down. That was it.. my wife predicted the downturn of Spencer Retail outlet. Unfortunately, the store manager did not take her seriously. Today, Spencer Retail has shut down the store. My wife did not use any analytics, any forecasting, nor any predictive tool... just her experience on a weekly basis to predict the store is going downhill.

Too often I have found inventory managers trying to order stocks basis individual movement of SKUs. So items that are moving slowly get eventually moved out of the merchandise portfolio for a store. No consideration is given to the fact that this item may actually complete the shopping basket of the customer. In the endeavour to focus on the fast moving SKUs, the slow moving ones are edged off the area of attention. This often leads to an incomplete shopping basket for the customer and forces her to visit other outlets to procure the missing items. When the number of missing items increases, the customer finds it convenient to avoid the store altogether and move to her regular kirana store or to another store outlet.

The need for inventory managers is to study the shopping basket to see if the slow moving item is part of a typical set of baskets. If so, then it would make sense to forecast the number of shopping baskets and then derive the number of SKUs to order basis the items in each shopping baskets.

This is not a very complex analysis. The purchase baskets are available from the POS units. Every purchase receipt is one basket. The first step is to group the purchase transactions into clusters of similarity. This can be done using either an cluster algorithm or an association algorithm. These two algorithm are significantly different and have their own objective. Once we know the number and contents of typical baskets, the next step is to forecast the number of baskets that will need to be filled. Basically, we are predicting the number of visitors to the store for each purchase baskets. When this is done, we can sum up the individual SKUs to come at the number of SKUs to stock so that all the purchase baskets are filled.

If only Spencers had done this analysis, my wife (and many others) would have continued to shop at the store and it would still be running today.

If you are in retail and want to know how this analysis can be done... get in touch with me at michaeldsilva@gmail.com

And before I windup, a reminder on the survey on the right side of the screen.

If you have come this far, do vote your view at the end of this post.