AI for Retail

CHAPTER 15AI for Merchandising

Current Roles, Processes, and Inefficiencies

The first step in retail is to decide what products a store should carry (SKU selection), in what quantities (shelf allocation), where it should go in the store (SKU position / space planning), and how much the retailer should charge (price). This process is colloquially called merchandising. Buyers build relationships with vendors to try to get the best deals on products they want to carry. Planners predict demand per SKU or per category based on last year's sales typically, and commit to orders for the year for specific products to all CPGs (consumer packaged goods) and private label vendors.

This gets pulled into a planogram (or POG) (see Figure 15.1).

This tells you what (SKU), should go where (position), and how much of it (# of facings). This is usually a very manual process. As an oversimplification, buyers select SKUs, and planners predict demand of those SKUs to help them plan where it will go in the store and how much of it. Since this is such a manual process, it happens one to four times a year, is performed on a national basis, perhaps a regional basis, and is usually never done on a per store basis as it would be impractical for merchants to go this low level and create a POG for every store, even though demand profiles are so wildly different on a per store level.

Every merchant agrees per‐store planograms would dramatically increase sales. In fact, the most successful merchant of all ...